Thioredoxin was initially identified by its ability to serve as an electron donor for ribonucleotide reductase in vitro. Whether it serves a similar function in vivo is unclear. In Saccharomyces cerevisiae, it was previously shown that ⌬trx1 ⌬trx2 mutants lacking the two genes for cytosolic thioredoxin have a slower growth rate because of a longer S phase, but the basis for S phase elongation was not identified. The hypothesis that S phase protraction was due to inefficient dNTP synthesis was investigated by measuring dNTP levels in asynchronous and synchronized wild-type and ⌬trx1 ⌬trx2 yeast. In contrast to wild-type cells, ⌬trx1 ⌬trx2 cells were unable to accumulate or maintain high levels of dNTPs when ␣-factor-or cdc15-arrested cells were allowed to reenter the cell cycle. At 80 min after release, when the fraction of cells in S phase was maximal, the dNTP pools in ⌬trx1 ⌬trx2 cells were 60% that of wild-type cells. The data suggest that, in the absence of thioredoxin, cells cannot support the high rate of dNTP synthesis required for efficient DNA synthesis during S phase. The results constitute in vivo evidence for thioredoxin being a physiologically relevant electron donor for ribonucleotide reductase during DNA precursor synthesis.Deoxyribonucleotide pools are carefully controlled in all cells to ensure efficient and yet accurate genome replication. The pools are not equimolar, but rather they have a characteristic asymmetry, with dGTP usually the smallest (1). For example, in Saccharomyces cerevisiae, the dGTP pool is 3-fold smaller than the dTTP pool (2, 3). Cells do not stockpile dNTPs, but rather they maintain them at levels just sufficient to support replication. For example, yeast contain fewer than 1.8 ϫ 10 6 dNTP molecules/cell (3), which is less than 20% of the amount minimally needed to copy the 1.2 ϫ 10 7 -bp yeast genome. Limiting dNTP pool size is important for replication fidelity. Artificial pool expansion results in increased mutagenesis, perhaps because of stimulation of chain extension from DNA mismatches at high dNTP concentrations (4). The dNTP pools are dynamic, increasing severalfold as eukaryotic cells enter S phase (3). For dNTP levels to remain elevated during S phase, cells must markedly boost their rate of dNTP synthesis to compensate for the rapid consumption of dNTPs at replication forks. Replicating cells are exquisitely sensitive to dNTP depletion. When dNTP synthesis is blocked by the addition of hydroxyurea, replication stops well before any dNTP pool is exhausted (5). Presumably, cells have evolved the mechanisms to limit dNTP accumulation to support replication fidelity and yet maintain dNTP levels above the minima required for efficient DNA replication and repair.A central enzyme controlling the rate of dNTP synthesis is ribonucleotide reductase (RNR), 3 a heterotetramer of two large and two small subunits. The enzyme is subject to extraordinary regulation. Its activity and substrate specificity are regulated by binding of different nucleoside triphosphates to two allo...