UmuD and RecA directly modulate the mutagenic potential of the Y family DNA polymerase DinB

Abstract

DinB is the only translesion Y family DNA polymerase conserved among bacteria, archaea, and eukaryotes. DinB and its orthologs possess a specialized lesion bypass function but also display potentially deleterious -1 frameshift mutagenic phenotypes when overproduced. We show that the DNA damage-inducible proteins UmuD(2) and RecA act in concert to modulate this mutagenic activity. Structural modeling suggests that the relatively open active site of DinB is enclosed by interaction with these proteins, thereby preventing the template bulging responsible for -1 frameshift mutagenesis. Intriguingly, residues that define the UmuD(2)-interacting surface on DinB statistically covary throughout evolution, suggesting a driving force for the maintenance of a regulatory protein-protein interaction at this site. Together, these observations indicate that proteins like RecA and UmuD(2) may be responsible for managing the mutagenic potential of DinB orthologs throughout evolution.

Figure 1. DinB interacts directly with UmuD₂ and RecA

A Farwestern blot demonstrates that UmuD directly interacts with ³²P-labeled (His)₆HMK-DinB. Either 50 or 100 pmols of UmuD or UmuD′ were separated by 12% SDS-PAGE and transferred to a PDVF membrane. The HMK-DinB protein probe was radioactively labeled and incubated with the membrane for 15 min. after which the membrane was exposed to film. B. Crosslinking experiment suggests that DinB interacts with the UmuD₂ homodimer. One hundred pmol of DinB and UmuD₂ were mixed in a 10 μL volume in 50 mM Hepes pH 7.5, 25–500 mM NaCl, and 1 mM DTT and incubated for 10 min at 25°C. C. DinB forms a stable binary and ternary complexes with RecA and UmuD₂. DinB(His)₆ pulls down RecA alone and in combination with UmuD and UmuD′ on a Ni²⁺ affinity resin. UmuDD′ cannot be observed to similarly interact.

Figure 2. UmuD₂ regulates the -1 frameshift activity of DinB in vivo

A. Lac+ reversion assay of the frameshift allele in CC108 demonstrates that UmuD₂ modulates DinB’s -1 frameshift function. B. Deletion of umuD⁺ results in a significant increase in -1 frameshift mutagenesis. A histogram of Lac⁺ mutants obtained from umuD⁺ and ΔumuD strains reveals that even the modest chromosomally encoded levels of UmuD₂ have a significant impact on -1 frameshift mutagenesis. Experiments were performed as indicated in materials and methods. Y-axis shows the frequency of plates with the number of Lac⁺ mutants indicated in the bins on the X-axis. C. UmuD₂ also affects the number of Lac⁺ revertants in an adaptive mutagenesis experiment. FC40 with empty vector (closed circles); FC40 with pUmuD′ (closed triangles); FC40 with pUmuD (open circles). Error bars represent one standard deviation.

Figure 3. Molecular characterization of the interaction between DinB and UmuD₂

A. Peptide array mapping of the UmuD binding interface on DinB reveals a surface composed of the thumb and finger domains of the polymerase. Several hydrophobic residues in the most strongly interacting peptide are conserved among DinB orthologs from organisms containing umuD. B. Peptide array mapping of the DinB binding interface on UmuD₂ reveals a discontinuous interface on a structural model of trans-UmuD₂ that is enlarged in an alternative isoenergetic trans-UmuD₂ conformer. C. Alanine mutants of DinB Phe172 or UmuD Asp 91 result in a weakened interaction determined by fluorescence spectroscopy. D. A low-copy number plasmid encoding DinB(F172A) (pYG768-F172A; open squares) is fully able to rescue the NFZ sensitivity of a ΔdinB E. coli strain bearing the pWSK29 empty vector (open circles) to the levels of an isogenic dinB⁺ strain (closed circles), just as a plasmid encoding wild-type DinB (pYG768; closed squares). Error bars represent one standard deviation determined from three independent experiments.

Figure 4. Single amino acid changes on the interface between DinB and UmuD perturb regulation of -1 frameshift activity

A. The DinB(F172A) variant has a lower affinity for UmuD₂ and is not as responsive as wild type DinB to regulation by UmuD₂. B. The UmuD(D91A) variant has a lower affinity for DinB and does not regulate -1 frameshift activity as well as wild type UmuD. Error bars represent one standard deviation.

Figure 5. UmuD₂ and RecA directly modulate DinB -1 frameshift function

A. Schematic of a mismatched DNA substrate that can be extended either by dGTP to generate a full length product or by dATP, thereby forming a dNTP stabilized misalignment and generating a -1 frameshift product. B. Plot of reaction velocity vs. dNTP substrate concentration for DinB in combination with RecA and UmuD₂. Extension of the GG mismatch with dATP by DinB in combination with stoichiometric quantities of RecA (open squares) is detectable but weaker than extension of a GC basepair in the same sequence context (open circles). The addition of saturating UmuD₂ (10 μM) profoundly inhibits DinB activity on a GG mismatch (closed squares) but stimulates DinB activity on a GC (closed circles) by more than 20-fold. Data for DinB alone (without RecA or UmuD₂) on the GG mismatch substrate are shown in open triangles. C. recA+ is required for UmuD dependent inhibition of DinB promoted -1 frameshift mutagenesis in vivo. Overeproduction of DinB promotes -1 frameshift mutagenesis in a ΔrecA background but the co-overproduction of UmuD₂ or a noncleavable UmuD(S60A) variant has little effect on mutation frequency. D. Plot of percent frameshift inhibition vs. UmuD variant concentration. The frameshift activity of DinB is efficiently inhibited by UmuD₂ (closed circles) but the frameshift activity of DinB(F172A) is more inert to UmuD₂ supression (open circles). The UmuD(D91A) variant is also very inefficient at inhibiting the -1 frameshift activity of wild-type DinB (open triangles). E. DinB efficiently inhibits UmuD₂ autocleavage in vitro. Wild-type DinB (open circles) inhibits UmuD₂ autocleavage far better than DinB(F172A) (closed circles). Error bars represent one standard deviation.

Figure 6. A TLS deficient variant of DinB is proficient for -1 frameshift function

A. The DinB(F13V) variant can promote -1 frameshift mutagenesis but is not controlled by co-overproduction of UmuD. B. The -1 frameshift activity of DinB(F13V) is poorly inhibited by UmuD₂ in vitro. Plot of frameshift activity vs. UmuD concentration indicates that DinB(F13V) (open squares) retains much of its frameshift activity at concentrations of UmuD that inhibit virtually all DinB frameshift activity (closed circles). All reactions contain RecA in stoichiometric ratios with DinB. Error bars represent one standard deviation.

Figure 7. RecA and UmuD₂ may enclose the open active site of DinB

A–B. In silico modeling of a ternary complex of the proteins. The surface representation of DinB is shown in blue, UmuD₂ in yellow, and RecA in orange. The DNA is relatively enclosed in the complex. C. Statistical covariance of DinB/pol κ residues across evolution. Residues that display statistical covariance with the UmuD₂ binding interface on E. coli DinB define an interface in a similar position on pol κ (shown in red), suggesting a possible rationale for the maintenance of this interface as a site of regulatory protein-protein interactions.