Stalled replication fork repair and misrepair during thymineless death in Escherichia coli

Abstract

Starvation for DNA precursor dTTP, known as 'thymineless death' (TLD), kills bacterial and eukaryotic cells alike. Despite numerous investigations, toxic mechanisms behind TLD remain unknown, although wrong nucleotide incorporation with subsequent excision dominates the explanations. We show that kinetics of TLD in Escherichia coli is not affected by mutations in DNA repair, ruling out excision after massive misincorporation as the cause of TLD. We found that the rate of DNA synthesis in thymine-starved cells decreases exponentially, indicating replication fork stalling. Processing of stalled replication forks by recombinational repair is known to fragment the chromosome, and we detect significant chromosomal fragmentation during TLD. Moreover, we report that, out of major recombinational repair functions, only inactivation of recF and recO relieves TLD, identifying the poisoning mechanism. Inactivation of recJ and rep has slight effect, while the recA, recBC, ruvABC, recG and uvrD mutations all accelerate TLD, identifying the protection mechanisms. Our epistatic analysis argues for two distinct pathways protecting against TLD: RecABCD/Ruv repairs the double-strand breaks, whereas UvrD counteracts RecAFO-catalyzed toxic single-strand gap processing.

Figure 1

Ideas about the possible trigger of thymineless death. Filled lines, template DNA strands; open lines, newly synthesized DNA strands; X, wrong DNA base. The ‘white death’ signs mark scenarios based on incorporation of wrong bases. The ‘black death’ sign marks the scenario based on replication fork stalling. (A) normal DNA duplex; (B) during dTTP starvation, wrong bases accumulate in DNA; (C) excision of wrong bases; (D) because of the lack of dTTP, excision repair is futile, leading to widening of excision gaps (lethal situation #1); (E) persistent single-strand gaps trigger recombination and formation of abnormal branched DNA species (lethal situation #2); (F) further widening of excision gaps leads to their overlapping and double-strand breaks (lethal situation #3); (G) normal replication fork; (H) replication fork stalling because of dTTP starvation, with unfinished Okazaki fragments accumulating; (I) improper recombination processing of a stalled replication fork (lethal situation #4).

Figure 2

Thymine starvation blocks DNA synthesis, induces SOS response and kills, but not through excision of wrong DNA bases. (A) Thymine starvation inhibits DNA synthesis. Normalization was to the rates at time 0.5 h without treatment. The values are means of three-to-six independent measurements ± SEM. Strains: thyA, KKW58; dnaB(Ts), KJK148. (B) Parameters of thymineless death (TLD) in a thyA mutant (KKW58) that have no other defects. The values are means of 37 independent measurements ± SEM (sometimes obscured by the symbols). (C) The kinetics of TLD in mutants, deficient in uracil excision, methyldirected mismatch removal or in nucleotide-excision repair. The wild-type TLD curve from panel B is shown for comparison. The values are means of three independent measurements ± SEM. Strains are: thyA ung, KJK78; thyA mutS, KJK179; thyA uvrA, KJK185. (D) Kinetics of SOS induction during TLD. The values are means of four independent measurements ± SEM. The strain is KJK106. As a control (white crosses), the same strain is treated with 100 ng/mL mitomycin C in the growth medium + thymidine for 1 h. After MC is removed, incubation is continued in the same medium.

Figure 3

Chromosomal fragmentation and genetics of SOS induction during thymine starvation. (A) A representative pulsed-field gel of chromosomal DNA of thyA and thyA recBCD mutants during thymine starvation. The strains are: wild type, KKW58; Δrec-BCD, KJK63. (B) Quantification of chromosomal fragmentation during thymineless death (TLD). The wild-type TLD curve from Fig. 2B is shown for comparison. The values are means of three independent measurements from gels like in ‘A’ ± SEM. (C) A scheme of the recombinational repair pathways. The SOS response is induced by RecA filament polymerization on the damaged chromosomes, licensed either by RecBC (at double-strand ends) or by RecFOR (at single-strand gaps). (D) The level of SOS induction in recombinational repair mutants. The measurement was taken four hours after thymidine removal. The data are means of four-to-seven independent measurements ± SEM. The strains are: Wild type, KJK106; recA, KJK125; recBCD, KJK129; recF, KJK127; recG, KJK135; ruvABC, KJK137; recBCD recF, KJK216; recG ruvABC, KJK217. (E) The kinetics of TLD in two types of lexA mutants. The lexA3 mutant produces an uncleavable SOS repressor and cannot be induced (IND−), while the lexA71 mutant does not produce the repressor and is constitutively induced for SOS. SulA is the SOS-induced cell division inhibitor and needs to be inactivated for lexA71 mutant to be viable. The wild-type TLD curve from Fig. 2B is shown for comparison. The values are means of four independent measurements ± SEM. Strais are: thyA, KKW58; thyA lexA3, KJK116; thyA lexA71, KJK123.

Figure 4

The roles of recombinational repair in thymineless death (TLD). In all panels, the wild-type curve from Fig. 2B is shown for comparison. (A) TLD kinetics in the recBCD and recD mutants. The values are means of five-to-eight independent measurements ± SEM. The strains are: recBCD, KJK63; recD, KJK99. (B) TLD kinetics in the recF, recO and recJ mutants. The values are means of 4–13 independent measurements ± SEM. The strains are: recF, KJK67; recO, KJK194; recJ, KJK192. (C) TLD kinetics in the recA mutant. The values are means of 24 independent measurements ± SEM. The strain is KJK61. (D) TLD kinetics in the ruvABC and recG mutants. The values are means of 7–11 independent measurements ± SEM. The strains are: ruvABC, KJK65; recG, KJK68.

Figure 5

Epistatic analysis of recombinational repair defects in thymineless death (TLD). In all panels, the single mutant TLD kinetics from Fig. 4 is shown for comparison. (A) TLD kinetics in the recBCD recF double mutant. The values are means of eight independent measurements ± SEM. The strain is KJK72. (B) TLD kinetics in the recG ruvABC double mutant. The values are means of seven independent measurements ± SEM. The strain is KJK73. (C) TLD kinetics in the recA recBCD double mutant. The values are means of five independent measurements ± SEM. The strain is KJK102. (D) TLD kinetics in the recA recF double mutant. The values are means of six independent measurements ± SEM. The strain is KJK104. (E) TLD kinetics in the recBCD ruvABC double mutant. The values are means of four independent measurements ± SEM. The strain is KJK212. (F) TLD kinetics in the recF ruvABC double mutant. The values are means of four to five independent measurements ± SEM. The strain is KJK211.

Figure 6

Thymineless death (TLD) in the rep and uvrD mutants. The wild-type and single mutant TLD kinetics from previous figures are shown for comparison. (A) TLD kinetics in the Δrep mutant. The values are means of three independent measurements ± SEM. The strain is KJK202. (B) TLD kinetics in the ΔuvrD mutant. The values are means of three independent measurements ± SEM. The strain is KJK112. (C) TLD kinetics in the uvrD recA double mutant. The values are means of three independent measurements ± SEM. The strain is KJK140. (D) TLD kinetics in the uvrD recF double mutant. The values are means of three independent measurements ± SEM. The strain is KJK144. (E) TLD kinetics in the uvrD recBCD double mutant. In contrast to all other TLD kinetics, for which strains were pre-grown in thymidine-supplemented medium at 28 °C, this set of strains was grown in the presence of thymidine at 42 °C to accelerate the otherwise impractically slow growth of the uvrD recBCD double mutant. The experiment itself was at 28 °C, as in all other cases. The values are means of three independent measurements ± SEM. The strain is KJK158. (F) Viability of recA, recBCD, recF and ruvABC mutants in the UvrD+ and uvrD mutant backgrounds. All strains are Thy+ and grown in LB. No value for the ruvABC uvrD double mutant reflects its known co-lethality.

Figure 7

A model of chromosomal lesion prevention, repair and misrepair during thymineless death. Old DNA strands are shown by filled lines, newly-synthesized strands are shown as open lines. Question mark, a hypothetical transition with an unclear mechanism.