Regulation of Mutagenic DNA Polymerase V Activation in Space and Time

Abstract

Spatial regulation is often encountered as a component of multi-tiered regulatory systems in eukaryotes, where processes are readily segregated by organelle boundaries. Well-characterized examples of spatial regulation are less common in bacteria. Low-fidelity DNA polymerase V (UmuD'2C) is produced in Escherichia coli as part of the bacterial SOS response to DNA damage. Due to the mutagenic potential of this enzyme, pol V activity is controlled by means of an elaborate regulatory system at transcriptional and posttranslational levels. Using single-molecule fluorescence microscopy to visualize UmuC inside living cells in space and time, we now show that pol V is also subject to a novel form of spatial regulation. After an initial delay (~ 45 min) post UV irradiation, UmuC is synthesized, but is not immediately activated. Instead, it is sequestered at the inner cell membrane. The release of UmuC into the cytosol requires the RecA* nucleoprotein filament-mediated cleavage of UmuD→UmuD'. Classic SOS damage response mutants either block [umuD(K97A)] or constitutively stimulate [recA(E38K)] UmuC release from the membrane. Foci of mutagenically active pol V Mut (UmuD'2C-RecA-ATP) formed in the cytosol after UV irradiation do not co-localize with pol III replisomes, suggesting a capacity to promote translesion DNA synthesis at lesions skipped over by DNA polymerase III. In effect, at least three molecular mechanisms limit the amount of time that pol V has to access DNA: (1) transcriptional and posttranslational regulation that initially keep the intracellular levels of pol V to a minimum; (2) spatial regulation via transient sequestration of UmuC at the membrane, which further delays pol V activation; and (3) the hydrolytic activity of a recently discovered pol V Mut ATPase function that limits active polymerase time on the chromosomal template.

Fig 1. Construction of a fluorescent UmuC reporter strain to visualize pol V regulation.

(A) Sketch depicting the regulation of pol V activity in E. coli. Transcriptional repression, targeted proteolysis, posttranslational modification and protein-protein interactions regulate levels of pol V Mut (UmuD’₂-UmuC-RecA-ATP), which is capable of mutagenic translesion DNA synthesis. The symbol “A” represents a molecule of ATP bound to RecA in the activated pol V Mut complex. (B) Replacement of the wild-type chromosomal umuC locus with umuC-mKate2 using λ_RED recombineering. (C) Western blots with anti-UmuC and anti-UmuD antibodies confirming the expression of full length UmuC-mKate2. Note that steady-state levels of the chimeric UmuC-mKate2 protein (lane ii, umuC-mKate2 background, strain RW1314) are approximately 20% of the wild-type UmuC protein (lane i, wild-type umuC background, RW574). Bands resulting from cross-reaction of the anti-UmuC antibody (see S1A Fig) with other proteins are also observed (marked with *) and indicate that similar amounts of lysate were loaded in each lane. Both strains are lexA(Def) recA(E38K) and thus constitutively express UmuD and UmuC and convert UmuD to UmuD'. Steady-state levels of UmuD' are similar in the wild-type umuC ⁺ (lane i) and umuC-mKate2 cells (lane ii). (D) In vivo mutagenesis assays performed on RW1314 show that UmuC-mKate2 is active for both spontaneous and damage-induced mutagenesis. Levels of mutagenesis are lower than that of the wild-type UmuC strain (RW574), but are consistent with overall lower steady state levels of the chimera.

Fig 2. Monitoring DNA damage induced expression of UmuC-mKate2 using time-lapse and time-sampling analysis.

(A) Time-lapse imaging of UmuC-mKate2 expressing cells (EAW282). Cells were grown at 37°C in flow cells and irradiated in situ with 10, 30 or 100 J.m^-2 of UV light (λ = 254 nm). Following irradiation fluorescence images were recorded every 5 min for 180 min. These images reveal that UmuC induction begins around 60 mins after DNA damage. (B) Quantification of the mean fluorescence signal of cell-containing regions within time-lapse image series after various does of UV-light. (C) Quantification of the mean number of molecules per cell and mean concentration of UmuC-mKate2 in time-sampling measurements after treatment with 30 J/m² UV light.

Fig 3. Changes in the cellular location of UmuC in response to UV irradiation.

(A) Montage of frames from a time-sampling movie showing UmuC-mKate2 diffusing along the cell periphery (EAW282). A peak-enhancing filter was applied to fluorescence images [57]. (B) Two-color super-resolution reconstruction showing co-localization of UmuC-mKate2 (red) and LacY-eYFP (blue) on the cell membrane (EAW191 containing pBAD-LacY-eYFP). (C) Individual cells cropped from time-lapse series showing UmuC transitioning from a membrane-associated to a cytosolic localization (EAW282). (D) Autocorrelation analysis of simulated images of E. coli cells displaying cytosolic and membrane-associated signals. Autocorrelation measurement can be conceptualised as follows: correlations are measured between an image and its distance-shifted duplicate. Correlations are measured as a function of shift distance (or distance lag). Autocorrelation analysis of cells displaying cytosolic signals produces a broad origin peak, whereas cells with membrane-associated signal produce a distinctive cross-peak (indicated with arrows). (E) Autocorrelation analysis of a simulated time-lapse series for an E. coli cell expressing a redistributing fluorescent protein. The simulation begins with no fluorescent protein signal. The protein is then induced, producing a membrane-associated signal before redistributing to the cytosol. Here the autocorrelation analysis is presented as a 2D contour plot. Blue areas indicate low correlation, whereas red areas indicate high correlation. Arrows indicate the membrane cross-peak that is visible when the cell has membrane-associated signal. (F) Similar autocorrelation analysis of experimentally acquired time-lapse series showing dose-dependent changes in UmuC localization (EAW282). At higher doses of UV the membrane cross-peaks begin to decline at ~90 min post-irradiation, while the origin peak becomes broader, indicating redistribution of UmuC-mKate2 into the cytosol. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis [57].

Fig 4. Visualization of UmuC redistribution by cross-correlation analysis and Western blotting.

(A) Cross-correlation analysis of LacY-eYFP and UmuC-mKate2 signals from time-lapse measurements (EAW191 containing pBAD-LacY-eYFP). Distance-dependent correlation values between LacY-eYFP and UmuC-mKate2 signals are presented as 2D contour plots. Blue areas indicate low correlation, whereas red areas indicate high correlation. Image pairs containing membrane-associated UmuC-mKate2 foci, thus co-localizing with membrane-defining LacY-eYFP signals, produce sharp cross-correlation peaks at lag = 0 and ± 0.8 μm. Image pairs containing cytosolic UmuC-mKate2 foci produce broader cross-correlation peaks at values between 0 and 0.8 μm. Red dotted lines mark the position of the secondary peaks in time and highlight a time-dependent trend towards lag values < 0.8 μm. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis. (B) UmuD, UmuD' and UmuC expressed from the recA ⁺ lexA ⁺ strain, RW118, after exposure to 30 Jm^-2 was monitored in whole cell extracts (WCE) and soluble fraction (SOL) at various time points after irradiation. The protein that is slightly smaller than UmuC and cross-reacts with the UmuC antibodies is indicated with an asterisk to the right of each lane. When detectible, UmuC is indicated by an arrow on the left of each lane. The UmuD and UmuD' proteins are indicated on the right hand side of the image

Fig 5. Post-synchronization of individual cells within time-lapse series (EAW282).

(A) Time-lapse images of individual cells post-synchronized to the point at which the intensity of UmuC-mKate2 signal reached a maximum (typically 90–150 min after treatment with 100 J/m² UV light). (B) Post-synchronized intensity vs time trajectories for 31 individual cells (of 100 total) that produced a single well-defined burst of UmuC-mKate2 synthesis. The remaining 69 cells either produced no detectable bursts of UmuC-mKate2 synthesis or produced multiple bursts of synthesis, preventing post-synchronization. (C) Mean intensity vs time trajectory over all cells showing three phases in the production of UmuC: (phase I) little UmuC-mKate2 produced; (phase II) increased production of UmuC-mKate2; (phase III) production of UmuC-mKate2 ceases and cellular levels decline. (D) Autocorrelation analysis of post-synchronized time-lapse series showing that the three phases in the production of UmuC correspond with changes in its cellular location: (phase I) weak membrane cross-peaks indicate low levels of membrane-associated UmuC-mKate2; (phase II) stronger membrane cross-peaks indicate production of membrane-associated UmuC-mKate2; (phase III) decreasing membrane cross-peaks and broadening of origin peak indicating redistribution of UmuC-mKate2 to the cytosol. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis [57].

Fig 6. Identifying the cellular location of UmuC-containing species using mutants with defects in the pol V activation pathway.

(A-D) Analysis of UmuC-mKate2 localization in strains with different genetic backgrounds; EAW282 (recA ⁺ lexA ⁺); EAW329 [recA ⁺ lexA ⁺ umuD(K97A)]; EAW307 (recA(E38K) lexA ⁺); and EAW455 [recA(E38K) lexA ⁺ umuD(K97A)]. For time-lapse/autocorrelation analysis, cells were treated at t = 0 min with UV light at 30 J/m². The images shown in time-lapse series are composites of bright-field images (cell outlines), raw fluorescence images and peak-filtered fluorescence images. The intensity range used for each channel was the same for all composites. The images represent stages in the UV-damage response (left to right: immediately after irradiation; peak signal—30 min; peak signal—15 min; peak signal; peak signal + 15 min; peak signal + 30 min). Autocorrelation analyses are again presented as a 2D contour plots, with the time course beginning at 0 min at the bottom. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis [57]. Side peaks in the autocorrelation function at -0.6 μm and 0.6 μm indicate membrane-associated UmuC-mKate2. More intense cross-peaks indicate a higher proportion of cells with membrane-localised distribution. A broad peak at the origin with no discernable membrane peaks is indicative of cytosolic UmuC-mKate2. Focus location maps (far right panels) were produced from analysis of shaking-culture cells. Cells were grown in EZ medium with glucose at 37°C. The entire 0.5 ml culture was irradiated with 10 J/m² UV light while sandwiched between two quartz plates then returned to shaking culture. Aliquots were taken every 10 min, placed on an APTES-treated coverslip, closed in with a second plain glass coverslip and imaged. 96 × 34 ms frames were recorded using 568 nm excitation light at a power of 1800 W/cm². All UmuC-mKate2 foci from all cells were mapped as Gaussians to a cell of standard size and shape. Maps for UV-irradiated cells include images recorded 10–120 min after irradiation.

Fig 7. Detection of UmuC location by immuno-electron microscopy and solubility in cell lysates.

(A–B) Immuno-electron microscopy of plasmid-expressed UmuC in the presence of umuD(C24D G25D) and umuD' in the E.coli strain RW1394. Sectioned electron-microscopy grids were incubated with anti-UmuC antibody and subsequently labeled with 30 nm gold beads conjugated to a secondary goat-anti-rabbit antibody. (A) Representative micrographs. Although the labeling density was low, gold beads were clearly visible for cells expressing UmuC, whereas control cells lacking UmuC showed virtually no beads. (B) Quantification of the proportion of beads associated with the membrane each sample. Beads found to be within 100 nm of the cell membrane were classed as membrane-associated. (C) Levels of soluble UmuC detected by western blotting (E. coli strain, TCH03). The two images are identical, except that upper panel is a darker exposure of the lower panel. UmuC was expressed alone, or co-expressed with UmuD' or UmuD(K97A). The images clearly show that UmuC is most soluble when co-expressed with UmuD' and least soluble when co-expressed with UmuD(K97A).

Fig 8. Colocalization of replisomes and mutasomes in wild-type (EAW282) and recAE38K cells (EAW307).

(A) Cross-correlation analysis of DnaX-YPet and UmuC-mKate2 signals from time-lapse measurements. Distance-dependent correlation between DnaX-YPet and UmuC-mKate2 signals are presented as 2D contour plots. Blue areas indicate low correlation, whereas red areas indicate high correlation. Image pairs containing co-localized replisome and mutasome foci produce high cross-correlation values over a short distance range. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis [57]. (B) Average projections of time-sampling movies showing replisome and mutasome foci. Arrows indicate co-localized replisome-mutasome pairs. A peak-enhancing filter was applied to fluorescence images prior to autocorrelation analysis [57]. (C-F) Changes in replisome and mutasome foci following 30 J/m² UV irradiation. (C) Number of mutasome foci per cell. (D) Proportion of mutasome foci that colocalize with a replisome focus. (E) Number of replisome foci per cell. (F) Proportion of replisome foci that colocalize with a mutasome focus.

Fig 9. Multiple mechanisms limit pol V activity on DNA.

(I) The activity of low fidelity pol V is regulated at the transcriptional level, with gene transcription and translation delayed by tight binding to the umuDC operator by LexA protein. (II) As documented here, activation of pol V is further delayed by sequestration of UmuC (and possibly UmuD) in the membrane. Release and final activation requires the conversion of UmuD to UmuD'. (III) Once activated, the length of time that pol V Mut is operative in TLS DNA replication is dependent on an ATPase activity intrinsic to the enzyme [14]. The symbol “A” represents a molecule of ATP bound to RecA in the activated pol V Mut complex.