Abstract | DNA replication in Xenopus embryos starts at random positions along the genome and at random times during S phase. This spatiotemporal randomness implies fluctuations in the completion times and can lead to cell death if replication takes longer than the cell cycle time (approximately 25 min). Surprisingly, while the typical completion time (approximately 20 min) is close to the cell cycle time, replication failure occurs only about 1 in 300 times. These observations raise an interesting question, known as the “random-completion problem”: How is replication timing accurately controlled despite the stochasticity? We use a nucleation-and-growth model and extreme-value statistics to address quantitatively the random-completion problem. We first show that Xenopus embryos solve the problem by using a large reservoir of potential replication origins that are increasingly likely to initiate, a situation that leads to robust control of replication timing. We also show that variations in the spatial distribution of origins have minimal effect on the accurate control of replication times. Finally, we show that replication in Xenopus minimizes, approximately, the number of proteins required for DNA synthesis. |