Saccharomyces cerevisiae is capable of de Novo pantothenic acid biosynthesis involving a novel pathway of beta-alanine production from spermine.

Pantothenic acid and beta-alanine are metabolic intermediates in coenzyme A biosynthesis. Using a functional screen in the yeast Saccharomyces cerevisiae, a putative amine oxidase, encoded by FMS1, was found to be rate-limiting for beta-alanine and pantothenic acid biosynthesis. Overexpression of FMS1 caused excess pantothenic acid to be excreted into the medium, whereas deletion mutants required beta-alanine or pantothenic acid for growth. Furthermore, yeast genes ECM31 and YIL145c, which both have structural homology to genes of the bacterial pantothenic acid pathway, were also required for pantothenic acid biosynthesis. The homology of FMS1 to FAD-containing amine oxidases and its role in beta-alanine biosynthesis suggested that its substrates are polyamines. Indeed, we found that all the enzymes of the polyamine pathway in yeast are necessary for beta-alanine biosynthesis; spe1Delta, spe2Delta, spe3Delta, and spe4Delta are all beta-alanine auxotrophs. Thus, contrary to previous reports, yeast is naturally capable of pantothenic acid biosynthesis, and the beta-alanine is derived from methionine via a pathway involving spermine. These findings should facilitate the identification of further enzymes and biochemical pathways involved in polyamine degradation and pantothenic acid biosynthesis in S. cerevisiae and raise questions about these pathways in other organisms.

the enzyme necessary for ␤-alanine biosynthesis in E. coli (1,2). Consistent with this, a structural homolog of aspartate-1decarboxylase is absent from the proteome of yeast (6), whereas structural homologs of all the other enzymes of the pantothenic acid pathway do exist in yeast. The gene ECM31, thought to be involved in cell wall maintenance (7), has homology to panB of E. coli and Aspergillus nidulans (8). The gene YIL145c is a panC ortholog, encoding pantothenate synthase, and has been shown to be functional in E. coli (9). The putative YHR063c gene has structural homology to panE, as noted in the Yeast Proteome Data base (6). Thus, the specific absence of a gene for aspartate-1-decarboxylase may appear to be consistent with the observation, first reported almost 60 years ago, that yeast require exogenous pantothenic acid for growth (4).
Decarboxylation of aspartate is not the only pathway for ␤-alanine biosynthesis. In some E. coli mutants, the source of ␤-alanine for pantothenic acid biosynthesis involves reduction of uracil to dihydrouracil followed by hydrolysis first to ␤-ureidoproprionate and second to CO 2 , NH 3 , and ␤-alanine (10). In addition, degradation of polyamines by amine oxidases can produce ␤-alanine (11,12), effectively making ␤-alanine from methionine (13). However, polyamine metabolism has never been implicated previously in pantothenic acid biosynthesis. We were therefore interested in the putative amine oxidase encoded by the yeast gene FMS1, which was originally identified as a multicopy suppressor of fen2 pantothenic acid import mutants and encodes a protein of 508 amino acids with sequence homology to FAD-containing amine oxidases (14). Pantothenic acid uptake deficiency in fen2 mutants causes CoA limitation, which affects yeast growth primarily by limiting ergosterol biosynthesis, suggesting a related role for FMS1 (5,14,15).
In this report we show that S. cerevisiae can synthesize ␤-alanine and is therefore capable of de novo biosynthesis of pantothenic acid. Furthermore, the biochemical pathway of ␤-alanine synthesis differs from that found in bacteria. We have found that ␤-alanine is formed from spermine via the amine oxidase encoded by FMS1. Thus, the ␤-alanine moiety of pantothenic acid is derived from methionine via S-adenosylmethionine and the polyamine pathway. These findings should facilitate the elucidation of other enzymes and metabolic intermediates involved in polyamine degradation and pantothenic acid biosynthesis and raise questions about these metabolic pathways in other organisms.

EXPERIMENTAL PROCEDURES
Yeast Strains and Media-The parental yeast strains BY4741 and BY4742 and their gene deletion derivatives (16) from the Saccharomyces Deletion Project were obtained through Research Genetics (Huntsville, AL); these strains are as follows: BY4741 (MATa his3 leu2 met15 ura3); BY4742 (MAT␣ his3 leu2 lys2 ura3); BY4741-0595 and BY4742-10595 (fms1⌬); BY4741-5757 and BY4742-15757 (fen2⌬); BY4741-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Multicopy Suppressor Screen-Yeast strain JHT14 was transformed with a yeast high copy library (21) and ϳ5 ϫ 10 4 Ura ϩ transformants were pooled, divided into aliquots, and stored in 25% glycerol at Ϫ70°C. Transformants were then spread at a density of 10 5 Ura ϩ cells per 10-cm Petri dish on synthetic agar medium containing 2% glucose, but lacking uracil, adenine, histidine, methionine, and pantothenic acid. After incubation for 3 days at 30°C, rapidly growing colonies occurred at a frequency of ϳ1.5 ϫ 10 Ϫ4 . The rapid growth phenotype of 37 of 55 colonies tested was found to be plasmid-dependent, based on the lack of growth on a selective medium containing 5-fluoroorotic acid and uracil (22). Ten of these plasmids were recovered from yeast by transformation of E. coli and confirmed to confer the rapid growth phenotype on selective medium when reintroduced into yeast strain JHT14. Based on comparison with sequence data in the Saccharomyces Genome Data base, six plasmids contained the ABZ1 locus, and four plasmids contained the FMS1 locus. The ABZ1 plasmids enhanced growth only because of the low concentration of paraaminobenzoic acid used in the medium, and they were not studied further.
Plasmids and DNA Manipulations-E. coli strains DH5␣ and DH10B (Life Technologies, Inc.) were used for DNA manipulations by standard methods (23). Plasmids were introduced into yeast using lithium acetate (24). Yeast DNA isolation for recovery of plasmids in E. coli was carried out using the Yeast Plasmid Isolation kit (Bio 101, Carlsbad, CA). The FMS1 coding sequence was amplified by polymerase chain reaction using oligonucleotide primers Fms1For-Xho (5Јccctcgagatgaatacagtttcaccag-3Ј) and Fms1Rev-Bam (5Ј-ttggatccctatttcagtaagtcag-3Ј) and ligated into the SalI and BamHI sites of YEp195AC (25) to create the ADH1-FMS1 overexpression vector. The E39Q substitution was made using QuikChange (Stratagene, La Jolla, CA) and mutagenic primers FmsE39Qupper (5Ј-gtcttgttcttcaggccagagatc-3Ј) and FmsE39Qlower (5Ј-gatctctggcctgaagaacaagac-3Ј). The DNA sequence of the entire mutant open reading frame was confirmed subsequently.
"Cross-feeding" Experiments-Log phase cultures of strain BY4742-10595 (fms1⌬) or strain BY4742-13316 (ecm31⌬) containing vector YEp195AC were prepared in synthetic medium lacking uracil and washed by centrifugation in water, and ϳ10 5 cells were spread on 10-cm Petri dishes containing synthetic agar medium lacking uracil and pantothenic acid. Log phase cultures of strain BY4742 and fen2⌬, fms1⌬, and ecm31⌬ deletion derivatives harboring either YEp195AC or the ADH1-FMS1 overexpression plasmid were prepared in synthetic medium lacking uracil, washed by centrifugation in water, and spotted onto the ecm31⌬ and fms1⌬ "lawns" at a density of ϳ10 7 cells per 5 l.
Plates were incubated at 30°C for 2 days, after which time "halos" of growing lawn cells formed around spots of cells that excreted pantothenic acid or downstream metabolites into the medium.

FMS1 Overexpression Enhances Growth in the Absence of
Pantothenic Acid-A high copy yeast genomic DNA library was screened for genes that enhanced growth in the absence of pantothenic acid, and plasmids containing genomic DNA in the region of the FMS1 locus were found. To determine whether FMS1, rather than other DNA sequences in these plasmids, was responsible for the enhanced growth, the FMS1 open reading frame was subcloned into an expression vector under the control of the ADH1 promoter and confirmed to have a DNA sequence identical to the published sequence (Ref. 14, Gen-Bank TM accession number X81848). This plasmid was introduced into yeast, and it was found that ADH1-FMS1, but not the empty vector, could enhance the growth of yeast on medium lacking pantothenic acid (Fig. 2).
FMS1, ECM31, and YIL145c Are Required for Pantothenic Acid Production in Yeast-As shown above, yeast grew well in the absence of pantothenic acid when FMS1 was overexpressed. This finding prompted us to test the currently available deletion strains, fms1⌬, ecm31⌬, YIL145c⌬, and fen2⌬ for pantothenic acid and ␤-alanine auxotrophy (see Fig. 1). The deletion strains and parental strain BY4742 were plated on medium lacking pantothenic acid and ␤-alanine or on medium supplemented with these compounds (Fig. 3A). The fms1⌬ strain required either pantothenic acid or ␤-alanine for growth. This is consistent with the results for overexpression of FMS1; overexpression of FMS1 enhanced growth in the absence of pantothenic acid/␤-alanine, whereas the fms1⌬ deletion totally blocked growth in the absence of pantothenic acid/␤-alanine.
In the same experiment, the ecm31⌬ and YIL145c⌬ strains could utilize pantothenic acid but differed from the fms1⌬ strain because they could not grow on ␤-alanine. Neither the fen2⌬ nor the parental strain required these supplements. Other potential metabolites, including ␤-ureidoproprionate, 5,6-dihydrouracil, l-aspartic acid, and 1,3-diaminopropane (100 M) did not support growth of the deletion strains (data not shown). The same results were also obtained using a different parental strain, BY4741, and its deletion derivatives (data not shown). Thus, based on the auxotrophic phenotypes, FMS1 is required for ␤-alanine production, whereas ECM31 and YIL145c are required downstream in the pantothenic acid pathway (Fig. 1).
Further evidence that FMS1 functions in the same pathway as ECM31 was obtained from a complementation analysis using the ADH1-FMS1 overexpression plasmid. This plasmid was introduced into fms1⌬, ecm31⌬, and fen2⌬ strains, which were then tested for growth in the absence of ␤-alanine (Fig. 3B). Growth occurred in the fms1⌬ and fen2⌬ strains, but not the ecm31⌬ strain, indicating that FMS1 is dependent on ECM31 for pantothenic acid biosynthesis. Thus, S. cerevisiae does not require exogenous pantothenic acid or ␤-alanine in the medium for growth, and FMS1 activity is rate-limiting for ␤-alanine biosynthesis under the conditions used.
Overexpression of FMS1 Results in Excretion of Excess Metabolites-Dramatically increased metabolic activity in the pantothenic acid pathway caused by FMS1 overexpression can be detected in cross-feeding experiments, in which cells excret-ing excess pantothenic acid cause halos of growth in lawns of pantothenic acid auxotrophs. Dense spots of wild-type, fen2⌬, fms1⌬, and ecm31⌬ strains containing either the ADH1-FMS1 overexpression vector or the empty vector were placed on lawns  3. FMS1 is necessary for ␤-alanine and pantothenic acid biosynthesis. A, parental BY4742, fen2⌬, fms1⌬, ecm31⌬, and YIL145c⌬ strains were plated (ϳ10 4 cells per spot) on synthetic complete medium lacking both ␤-alanine and pantothenic acid or on medium supplemented with either of these compounds (0.1 mM), as indicated, and incubated at 30°C for 5 days. B, YEp195AC empty vector (sectors 1-3) or the ADH1-FMS1 overexpression plasmid (sectors 4 -6) was introduced into the fen2⌬ (sectors 1 and 4), ecm31⌬ (sectors 2 and 5), or fms1⌬ (sectors 3 and 6) strains (BY4742 derivatives), streaked on uracil drop-out medium with or without ␤-alanine (0.1 mM), as indicated, and incubated for 3 days at 30°C. of either ecm31⌬ or fms1⌬ cells on medium lacking pantothenic acid (Fig. 4). After incubation, halos formed around each of the FMS1-overexpressing strains, with the exception of the ecm31⌬ strain, on both lawns. No halos formed around strains harboring empty vector. The simplest explanation for the halos is that FMS1 overexpression results in excretion of excess pantothenic acid, which is then taken up by the lawn cells, allowing them to grow. Overexpression of FMS1 in the ecm31⌬ strain did not result in a halo, further confirming that FMS1 and ECM31 function in the same pathway.
The FAD-binding Domain of Fms1p Is Necessary for ␤-Alanine Production-Although it is required for ␤-alanine production, FMS1 encodes a protein that has no structural homology to bacterial aspartate-1-decarboxylases. Instead, Fms1p has homology to FAD-containing amine oxidases (14) and likewise contains a GXGXXG dinucleotide-binding motif similar, for example, to Candida albicans Cbp1p, human monoamine oxidases A and B, and the peroxisomal acetylspermidine oxidase Aso1p of Candida boidinii (Fig. 5A). To assess the role of FAD in ␤-alanine production, we made the E39Q substitution mutant, equivalent to the substitution that was shown to abolish FAD binding and catalytic activity of monoamine oxidase B (26 -28). The resulting ADH1-fms1(E39Q) expression plasmid was introduced into the fms1⌬ strain, and transformants were tested for growth in the absence of ␤-alanine and pantothenic acid (Fig. 5B). The E39Q mutant did not complement the ␤-alanine and pantothenic acid auxotrophy of the fms1⌬ strain, consistent with a role for FAD in the mechanism of ␤-alanine production.
FMS1 Links Polyamine Biosynthesis with Pantothenic Acid Production in Yeast-The sequence homology of Fms1p to amine oxidases and the apparent role of FAD in ␤-alanine production by Fms1p suggested that polyamines could provide the substrates for Fms1p. We therefore tested deletion mutants of the polyamine pathway (29,30) for ␤-alanine auxotrophy. Parental, spe1⌬, spe2⌬, spe3⌬, spe4⌬, fms1⌬, and ecm31⌬ strains were plated on medium lacking pantothenic acid, ␤-alanine, and polyamines or on medium supplemented with pantothenic acid, ␤-alanine, spermine, spermidine, or putrescine (Fig. 6). All four of the spe⌬ mutants were able to grow when FIG. 4. FMS1 overexpression leads to excretion of pantothenic acid. Lawns of ecm31⌬ and fms1⌬, containing YEp195AC empty vector, were spread on uracil drop-out medium lacking ␤-alanine and pantothenic acid. Dense spots of fms1⌬, ecm31⌬, fen2⌬, or parental BY4742 harboring YEp195AC empty vector or the ADH1-FMS1 overexpression vector were placed onto the lawns, as indicated, followed by incubation at 30°C for 3 days. one of the compounds spermine, ␤-alanine, or pantothenic acid was added to the medium. In addition, spe1⌬ and spe3⌬ could also grow on spermidine, and spe1⌬ could grow on putrescine. As expected, the fms1⌬ and ecm31⌬ strains could grow on pantothenic acid but could not utilize any of the polyamine compounds. Thus, biosynthesis of ␤-alanine and pantothenic acid is dependent on the polyamine biosynthetic pathway, consistent with production of ␤-alanine via polyamine degradation (11)(12)(13). The source of the carbon atoms in ␤-alanine would therefore be from methionine via spermine. Based on these results, the relationship of the polyamine pathway to pantothenic acid biosynthesis and the key genes involved are illustrated in Fig. 7. The spe⌬ strains were not able to grow on the potential polyamine degradation metabolite 1,3-diaminopropane (data not shown), apparently ruling out this compound as an intermediate in ␤-alanine biosynthesis. DISCUSSION S. cerevisiae Can Synthesize ␤-Alanine and Pantothenic Acid-Yeast have been reported to require a supplement of either pantothenic acid or ␤-alanine, from which it has been inferred that they cannot synthesize pantothenic acid de novo (4,5). Contrary to this, we found that overexpression of the yeast gene FMS1, encoding a putative amine oxidase, allowed strong growth on medium lacking pantothenic acid. Furthermore, when FMS1 was overexpressed under the control of the ADH1 promoter, excess pantothenic acid was excreted from the cells. Thus, yeast clearly have the capacity to synthesize pantothenic acid de novo when FMS1 is overexpressed. To eliminate the possibility that the pantothenic acid biosynthesis was a meta-bolic abnormality caused by FMS1 overexpression, we analyzed gene deletion mutants. The fms1⌬ strains were auxotrophic for ␤-alanine and could grow when supplemented with either ␤-alanine or pantothenic acid. Deletions in genes that have structural homology to the bacterial genes of the pantothenic acid pathway, ECM31 and YIL145c (see Fig. 1), caused pantothenic acid auxotrophy, but these strains did not grow on a ␤-alanine supplement, indicating that these genes are downstream in the pathway (see Fig. 1). These auxotrophic phenotypes indicate that FMS1, ECM31, and YIL145c are normally involved in pantothenic acid biosynthesis. Direct evidence that FMS1 and ECM31 are in the same pathway came from the finding that FMS1 requires ECM31 activity to make pantothenic acid; the ecm31⌬ deletion eliminated both the ability of FMS1 to enhance growth in the absence of pantothenic acid and also eliminated its ability to cause pantothenic acid excre-  Fig. 6. The dashed arrow between spermine and ␤-alanine represents the novel connection between polyamine degradation and pantothenic acid biosynthesis in yeast, involving FMS1 and additional unidentified genes. B, chemical structures for methionine, decarboxyadenosylmethionine (dcAdoMet), putrescine, spermidine, spermine, 3-aminopropanal, ␤-alanine, and (R)-pantothenate. The bold lines represent the three-carbon moiety that is derived from methionine in yeast. tion. Thus, pantothenic acid biosynthesis is a natural part of metabolism in S. cerevisiae, and production of the ␤-alanine required involves a putative amine oxidase, encoded by FMS1.
Spermine Is Required for ␤-Alanine Biosynthesis in Yeast-Three different enzymatic pathways have been shown to produce ␤-alanine: decarboxylation of aspartate (1,2), degradation of pyrimidines (10), and degradation of polyamines (13). The polyamine pathway has not been implicated previously in pantothenic acid biosynthesis, and spermine, in particular, has no previously identified physiological function in yeast (31). The involvement of the polyamine pathway is suggested by the Fms1p amino acid sequence, which has structural homology to FAD-containing amine oxidases (14), some of which are involved in the oxidative degradation of polyamines (4). In addition, we showed that a FAD binding site mutant of FMS1, E39Q, did not complement the fms1⌬ mutant, consistent with a role for oxidation by the Fms1p protein. We therefore investigated deletion mutants of the polyamine pathway to see whether this pathway is required for ␤-alanine synthesis. Indeed, spe1⌬, spe2⌬, spe3⌬, and spe4⌬ mutants were all auxotrophic for ␤-alanine on a medium that lacked polyamines. This showed that synthesis of spermine is required for ␤-alanine biosynthesis in yeast. A more detailed analysis of which polyamine pathway intermediates could support growth of the spe⌬ and fms1⌬ mutants confirmed this conclusion (Fig. 6). Thus, in yeast, ␤-alanine is derived from methionine via spermine, making polyamine degradation part of pantothenic acid biosynthesis (Fig. 7).
We found that the auxotrophic phenotypes of the spe⌬ mutants for polyamines were readily observable on minimal synthetic medium in the absence of ␤-alanine or pantothenic acid (Fig. 6). This contrasts with previous reports, in which special precautions in medium preparation were required, such as HCl washing of glassware and avoidance of autoclave use, to eliminate contaminating amines, and in which many cell divisions were required to deplete intracellular pools of polyamines (32). The difference is simply in the presence or absence of ␤-alanine or pantothenic acid; in their absence, a relatively high level of polyamine metabolism is required to meet ␤-alanine requirements, such that contaminated glassware and intracellular pools do not make a significant contribution. In the presence of ␤-alanine or pantothenic acid, as customary in yeast media, low levels of contaminating polyamines are sufficient for essential processes, such as hypusine synthesis (33), which are unrelated to the pantothenic acid pathway.
The Pathway from Spermine to ␤-Alanine-Amine oxidases have been shown to catalyze a number of different degradation reactions for polyamines, producing various aldehydes and amines, such as 3-aminopropanal and 1,3-diaminopropane, respectively (13). Thus, the simplest hypothesis is that the Fms1p enzyme converts spermine to 3-aminopropanal and spermidine and that aldehyde dehydrogenases, for which there are seven genes in yeast (34), would be required to convert the 3-aminopropanal to ␤-alanine. A less direct route between spermine and ␤-alanine could, in principle, involve the intermediate 1,3-diaminopropane (13). However, this compound was not able to support growth of the spe⌬ mutants in the absence of ␤-alanine and therefore appears not to be on the pathway in yeast. The simple phenotype of ␤-alanine auxotrophy in yeast will help identify the metabolic intermediates and additional enzymes involved.
Regulation of FMS1 Activity-It may seem unexpected that yeast would have the capacity to make pantothenic acid and yet require a supplement for efficient growth on customary yeast media. In fact, on medium containing glycerol or acetate as the sole carbon source, we found that pantothenic acid and ␤-ala-nine were not rate-limiting for growth (data not shown). Thus, FMS1 activity is growth-limiting only on glucose medium. This simple observation may explain the carbon source-dependent phenotype (catabolite repression) reported for the fen2 pantothenate transporter mutant (15). In the light of the finding that pantothenic acid biosynthesis is a natural part of yeast metabolism, we propose that growth of the fen2 mutant, which cannot absorb pantothenic acid from the medium, depends on pantothenic acid synthesis inside the yeast cells. The fen2 mutant would therefore be growth-limited on glucose because of insufficient FMS1 expression caused by the presence of glucose. Likewise, wild-type strains would depend on internal synthesis when pantothenic acid was absent from the medium and would be growth-limited by insufficient FMS1 expression on glucose medium. These observations suggest that FMS1 activity is regulated and raise questions concerning the mechanism of regulation of the pantothenic acid pathway in yeast.
␤-Alanine Biosynthesis in Other Organisms-The finding that ␤-alanine biosynthesis is different between yeast and bacteria raises questions as to how other organisms, such as fungi and plants, make ␤-alanine. At the present time in the public sequence data bases there are over a dozen identifiable aspartate-1-decarboxylase genes from different prokaryotic species, whereas this enzyme does not appear to be present in eukaryotic species. In contrast, proteins of significant sequence homology to Fms1p can be found in eukaryotes, in particular in plants, but not in prokaryotes. The closest sequence similarity to Fms1p is in Cbp1p from the yeast C. albicans, a protein with steroid binding activity (14,35). This suggests that plants and lower eukaryotes generally produce ␤-alanine and hence CoA by a polyamine degradation pathway, as described here for yeast.