DNA polymerase IV primarily operates outside of DNA replication forks in Escherichia coli

Abstract

In Escherichia coli, damage to the chromosomal DNA induces the SOS response, setting in motion a series of different DNA repair and damage tolerance pathways. DNA polymerase IV (pol IV) is one of three specialised DNA polymerases called into action during the SOS response to help cells tolerate certain types of DNA damage. The canonical view in the field is that pol IV primarily acts at replisomes that have stalled on the damaged DNA template. However, the results of several studies indicate that pol IV also acts on other substrates, including single-stranded DNA gaps left behind replisomes that re-initiate replication downstream of a lesion, stalled transcription complexes and recombination intermediates. In this study, we use single-molecule time-lapse microscopy to directly visualize fluorescently labelled pol IV in live cells. We treat cells with the DNA-damaging antibiotic ciprofloxacin, Methylmethane sulfonate (MMS) or ultraviolet light and measure changes in pol IV concentrations and cellular locations through time. We observe that only 5-10% of foci induced by DNA damage form close to replisomes, suggesting that pol IV predominantly carries out non-replisomal functions. The minority of foci that do form close to replisomes exhibit a broad distribution of colocalisation distances, consistent with a significant proportion of pol IV molecules carrying out postreplicative TLS in gaps behind the replisome. Interestingly, the proportion of pol IV foci that form close to replisomes drops dramatically in the period 90-180 min after treatment, despite pol IV concentrations remaining relatively constant. In an SOS-constitutive mutant that expresses high levels of pol IV, few foci are observed in the absence of damage, indicating that within cells access of pol IV to DNA is dependent on the presence of damage, as opposed to concentration-driven competition for binding sites.

Fig 1. Construction of E. coli cells expressing labelled pol IV and analysis of bypass and mutagenic functions.

(A) Construction of EAW633 (lexA⁺ dinB-YPet). The dinB gene of E. coli K12 MG1655 was modified using λ_RED recombineering so that pol IV is expressed as a fusion with the bright yellow fluorescent protein YPet (DinB-YPet). (B) Confirmation of DinB-YPet expression in ciprofloxacin-treated cells. (Upper part) Western blot of extracts from cells (treated with 30 ng/ml ciprofloxacin for 120 min), developed using anti-DinB antibodies. Lanes: i) molecular weight marker; ii) MG1655; iii) EAW633 (dinB-YPet); iv) EAW643 (dinB-YPet dnaX-mKate2); v) BL21 pLysS pET-DinB (uninduced cell extract). Bands corresponding to full length DinB-YPet are clearly visible in lanes ii and iii. A small amount of two DinB-containing fragments are also visible. Fragment 1 corresponds to DinB+linker. Fragment 2 corresponds to DinB +/- one or two residues. (Lower part) Results of densitometry measurements for lanes ii–iv. DinB-YPet is expressed at levels equivalent to wild-type DinB, however ~20% is proteolysed within the cells. (C) DinB-YPet retains lesion bypass activity. Strains were grown to exponential growth phase (OD₆₀₀ = 0.2), serial diluted, and spotted onto LB agar plates containing 8 μM of 4-nitroquinolone-1-oxide (NQO). Because of an inability to bypass lesions induced by NQO, cells lacking dinB are sensitized by 3 orders of magnitude relative to wild type cells. Cells expressing DinB-YPet survival to levels equivalent to wild-type cells, indicating that DinB-YPet retains full lesion bypass activity. (D) DinB-YPet facilitates mutation to ciprofloxacin resistance. Approximately 10⁸ log-phase cells were spread onto LB agar plates containing 40 ng/ml ciprofloxacin and incubated at 37°C for 13 days. Colonies appearing on the plates were counted on days 4, 8 and 13. The number of new colonies appearing between each interval was determined and normalised against viable cell counts, as described in reference [9]. Cells lacking dinB produced only 10% as many ciprofloxacin-resistant colonies as wild-type cells. DinB-YPet expressing cells produced similar number of resistant colonies as wild-type cells, indicating that DinB-YPet supports mutagenic pol IV activities.

Fig 2. Single-molecule imaging of cells expressing DinB-YPet.

(A) Detection of DNA-bound molecules in single-molecule images. Molecules of DinB-YPet that are not bound to DNA will diffuse quickly (D ≈ 10⁵ μm²/s for a typical cytosolic protein) and thus signals from individual molecules will blur over the entire cell in our images (exposure time = 30–100 ms). Molecules of DinB-YPet that are bound to DNA, however, experience greatly reduced motion and thus appear as punctate foci. Because of this diffusional contrast, it is possible to detect individual molecules of DinB-YPet when bound to DNA. (B) Single-molecule sensitive fluorescence image of undamaged EAW643 cells showing low-level DinB-YPet signals at 50 ms exposure time. (C) Average projection of rapid acquisition (effective exposure time 306 ms) for DinB-YPet (left) and DinBD103N-YPet (right). (D)Time-lapse imaging of pol IV up-regulation in response to ciprofloxacin treatment. Images shown are a montage of a three-hour time-lapse series. Cells were initially grown in rich medium in the absence of exogenous DNA damage. At t = 0 min, the flow cell inlet was switched to medium containing 30 ng/ml ciprofloxacin. At each field-of-view, a bright-field image and a DinB-YPet fluorescence image were collected every 5 min for 180 min. Time stamps indicate hours after ciprofloxacin addition.

Fig 3. Scatter plots of cell-size and fluorescence signal parameters from time-lapse imaging of DinB-YPet cells treated with ciprofloxacin.

White points indicate individual data-points, while blue-to-red contours indicate frequencies of observations. Blue areas indicate regions of the plot containing few data points; red areas indicate regions containing a large number of data points. Frequencies were normalised at each time-point to the maximum value at that time-point. (A) Distribution of cell lengths based on bright-field images, showing ciprofloxacin-induced filamentation. (B) DinB-YPet fluorescence per cell, measured as the mean pixel intensity within each cell, showing up-regulation of DinB-YPet. (C) Density of DinB-YPet foci, measured as the number of foci per cell area (μm²), showing the density remains relatively constant over the three-hour measurement. (D) Integrated fluorescence intensity of foci. Each focus was fit with a 2D Gaussian function; the volume under this function represents the integrated fluorescence intensity. Foci become brighter over the course of the measurement, indicating that a higher number of DinB-YPet molecules bind at each binding site. We conservatively estimate that >100 cells were used in each measurement.

Fig 4. Colocalisation of pol IV with replisomes.

(A) Montage of two-colour time-lapse movie recorded after treatment 30 ng/mL ciprofloxacin. Pol IV (DinB-YPet) foci appear green and replisome (DnaX-mKate2) foci appear in magenta. Colocalised foci appear white. For display purposes, images were subjected to spatial filtering to enhance foci [47]. (B-C) Analysis of colocalisation distances for foci detected in two-colour images. (B) Diagram of area shells used for colocalisation analysis. As colocalisation is a radial measurement, histograms of colocalisation distances are constructed using bins of linearly increasing area rather than distance. (C) Histograms of colocalisation distances for foci within a doubly labelled replisome strain (EAW203; dnaX-YPet dnaQ-mKate2) and a two-colour pol IV/replisome strain (EAW643; dinB-YPet dnaX-mKate2). As expected, distances between DnaX-YPet and DnaQ-mKate2 foci fall within a narrow distribution, indicative of ‘tight’ colocalisation. Distances between DinB-YPet and DnaX-mKate2 foci present a much broader distribution, indicative of ‘loose’ colocalisation. (D) Plot of the number of pol IV and replisome foci per EAW643 cell as a function of time. Data were compiled from ten technical replicates. Shaded areas indicate the standard error of the mean between these replicates. Some cells were lost from the coverslip surface during the measurement. A total of 188 cells remained bound and were analysed over the full course of the measurement. (E) Plots of mean cell intensity and colocalisation between pol IV and replisome foci. The mean cell intensity (grey shaded area) is a direct measure of the pol IV concentration in cells. Foci located within 200 nm of each other were defined as being colocalised. Colocalisation was measured in two ways: the proportion of pol IV foci that contain a colocalised replisome focus (black line), and the proportion of replisome foci that contain a colocalised pol IV focus (orange line). Data were compiled from ten technical replicates. Shaded areas indicate the standard error of the mean between these replicates. The total number of cells analysed were not determined in these measurements. We conservatively estimate that >1000 cells were used in each measurement. The analysis includes a total of 17005 DnaX-mKate2 foci and 12408 DinB-YPet foci.

Fig 5. Colocalisation of pol IV with replisomes in the presence of additional fluorescently labelled pol IV expressed from a plasmid.

(A) Mean cell intensity measurements for EAW643 cells (Pol IV⁺; light grey line) and EAW643 cells containing pPFB1188 (expressing additional DinB-eYFP from the dinB promoter; Pol IV⁺⁺; dark grey line). Cells containing pFB1188 clearly express much higher levels of labelled pol IV, however because cells contain an unknown ratio two different YFPs (DinB-YPet and DinB-eYFP), it is not possible to measure the pol IV concentration. (B) Plots of mean cell intensity and colocalisation between pol IV (DinB-YPet/DinB-eYFP) foci and replisome (DnaX-mKate2) foci. The mean cell intensity (grey shaded area) is a convoluted measure of the combined DinB-YPet and DinB-eYFP concentrations in cells. Colocalisation was measured in two ways: the proportion of pol IV foci that contain a colocalised replisome focus (black line), and the proportion of replisome foci that contain a colocalised pol IV focus (orange line). Data were compiled from ten technical replicates. The total number of cells analysed were not determined in these measurements. We conservatively estimate that >500 cells were used in each measurement. The analysis includes a total of 27651 DnaX-mKate2 foci and 31978 DinB-YPet/DinB-eYFP foci.

Fig 6. Comparison of DinB-YPet behaviour in untreated, lexA(Def) cells and ciprofloxacin-treated lexA⁺ cells.

(A) Representative images of ciprofloxacin-treated lexA⁺ cells (left) and untreated lexA(Def) cells (right). (B) Representative intensity vs time trajectories for DinB-YPet signals in the vicinity of replisomes. Additional, randomly selected trajectories appear in S7 Fig (ciprofloxacin-treated lexA⁺ cells) and S8 Fig (untreated lexA(Def) cells). 5×5 pixel regions of interest were placed at replisome foci, then used to monitor fluctuations in DinB-YPet signals (see panel A). In ciprofloxacin-treated lexA⁺ cells, DinB-YPet signals are elevated in the vicinity of replisomes for multiple frames, indicating long-lived binding events. In untreated lexA(Def) cells no events are visible in which the DinB-YPet is elevated in the vicinity of replisomes for more than a single 34 ms frame, indicating no long-lived binding events.

Fig 7. Timeline of translesion DNA synthesis based on single-molecule imaging studies.

Pol IV is expressed relatively early after DNA damage is incurred and is allowed access to replisomes until cells abruptly transition into the late stage. At this transition, pol IV is ejected from replisomes and a second TLS polymerase, pol V Mut becomes activated. Pol IV continues to act on non-replisome substrates. The timescale indicated for these transitions is likely to be specific to our growth conditions (EZ glucose medium; APTES-treated flow cell; 37°C). We anticipate that under different conditions the same transitions would be observed, but at different time-points.