The translesion DNA polymerases Pol ζ and Rev1 are activated independently of PCNA ubiquitination upon UV radiation in mutants of DNA polymerase δ

Abstract

Replicative DNA polymerases cannot insert efficiently nucleotides at sites of base lesions. This function is taken over by specialized translesion DNA synthesis (TLS) polymerases to allow DNA replication completion in the presence of DNA damage. In eukaryotes, Rad6- and Rad18-mediated PCNA ubiquitination at lysine 164 promotes recruitment of TLS polymerases, allowing cells to efficiently cope with DNA damage. However, several studies showed that TLS polymerases can be recruited also in the absence of PCNA ubiquitination. We hypothesized that the stability of the interactions between DNA polymerase δ (Pol δ) subunits and/or between Pol δ and PCNA at the primer/template junction is a crucial factor to determine the requirement of PCNA ubiquitination. To test this hypothesis, we used a structural mutant of Pol δ in which the interaction between Pol3 and Pol31 is inhibited. We found that in yeast, rad18Δ-associated UV hypersensitivity is suppressed by pol3-ct, a mutant allele of the POL3 gene that encodes the catalytic subunit of replicative Pol δ. pol3-ct suppressor effect was specifically dependent on the Rev1 and Pol ζ TLS polymerases. This result strongly suggests that TLS polymerases could rely much less on PCNA ubiquitination when Pol δ interaction with PCNA is partially compromised by mutations. In agreement with this model, we found that the pol3-FI allele suppressed rad18Δ-associated UV sensitivity as observed for pol3-ct. This POL3 allele carries mutations within a putative PCNA Interacting Peptide (PIP) motif. We then provided molecular and genetic evidence that this motif could contribute to Pol δ-PCNA interaction indirectly, although it is not a bona fide PIP. Overall, our results suggest that the primary role of PCNA ubiquitination is to allow TLS polymerases to outcompete Pol δ for PCNA access upon DNA damage.

Fig 1. pol3-ct increases UV resistance of yeast cells deficient for PCNA ubiquitination independently of HR.

Survival curves of haploid cells deficient for PCNA ubiquitination after exposure to UV light. Red arrows highlight increased UV resistance in the presence of pol3-ct. (A) Suppression of rad18Δ cells UV hypersensitivity by the pol3-ct allele. (B and C) The increased UV resistance of rad18Δ pol3-ct cells is HR-independent. (D) pol3-ct suppresses UV sensitivity of HR-deficient pol30-K164R cells.

Fig 2. pol3-ct increases UV resistance in the absence of PCNA polyubiquitination.

Survival curves of haploid cells deficient for PCNA poly-ubiquitination after exposure to UV light. Red arrows highlight increased UV resistance in the presence of pol3-ct. (A and C) Suppression of UV hypersensitivity in rad5Δ cells by the pol3-ct allele in the presence or absence of Rad51. (B and D) Suppression of mms2Δ-associated UV sensitivity by the pol3-ct allele in the presence or absence of Rad51. The weaker suppression of mms2Δ UV sensitivity by the pol3-ct allele is shown with a smaller and dotted red arrow.

Fig 3. pol3-ct effects in cells lacking TLS polymerases.

Survival curves of TLS-deficient haploid cells after exposure to UV light. (A) pol3-ct increases the UV resistance of rad30Δ cells (red arrow) only in the presence of Rad51. (B) pol3-ct does not increase UV resistance of rev3Δ cells in the presence of Rad51. A minor pol3-ct effect is observed in the absence of Rad51. (C) pol3-ct does not restore UV resistance of rev1Δ cells in the presence Rad51. A minor pol3-ct effect is observed in the absence of Rad51.

Fig 4. Suppression of rad18Δ UV phenotypes by pol3-ct requires REV3 and REV1.

Survival curves of UV-irradiated haploid cells with defective PCNA ubiquitination and TLS. (A) Suppression of rad18Δ UV sensitivity by pol3-ct (red arrow) does not require RAD30. (B) Suppression of rad18Δ UV sensitivity by pol3-ct requires REV3. (C) Suppression of rad18Δ UV sensitivity by pol3-ct requires REV1. (D) pol3-ct does not suppress the effect of the rev1-12 allele. rev1Δ cells are complemented with a plasmid containing either the WT REV1 gene (pREV1) or the rev1-12 allele (prev1-12). (E) Suppression of rad18Δ UV sensitivity by pol3-ct (red arrow) does not require a functional UBM domain in Rev1. rad18Δ rev1Δ cells are complemented with a plasmid containing either the WT REV1 gene (pREV1) or the rev1-12 allele (prev1-12). (F) Increase of UV-induced mutagenesis in rad18Δ cells carrying pol3-ct. Each dot represents the can^R frequency measured in one experiment. The median value for each strain is represented by the horizontal bar. For this experiment, the rad18Δ and rad18Δ pol3-ct strains were isolated from two independent progenies (#1 and #2), and in both cases the can^R frequency between rad18Δ and rad18Δ pol3-ct was significantly different (Mann-Whitney test): **, P = 0.0043 (#1) and *, P = 0.0353 (#2).

Fig 5. Functional analysis of the interaction between Pol3-CTD and Pol31 in cell survival after UV irradiation.

(A) Structural model of the Pol3-CTD/Pol31/P66 complex. Pol3 CTD contains two conserved cysteine-rich metal-binding motifs (CysA and CysB) that are separated by two α-helices. The Arg1043 residue of Pol3 lies within helix α2 between the Pol3 LSKW residues and Pol31. In addition, Pol31 Asp304 is very close to Pol31 interacting surface opposite to Pol3 Arg1043. The position of the LSKW motif at the C-terminal end of Pol 3 CTD is indicated by Pol3-LSKW. The conserved L1094 and W1097 residues of this motif are indicated by sticks. This motif is deleted in the pol3-ct mutant. The Pol3 residue R1043 within helix α2 is also indicated by sticks. The Pol3 residues F1002 and I1003 are located five amino-acids upstream of C1009 (the first amino-acid included in the structural model). These five amino-acids are indicated by black dots to show the proximity of Pol3-F1002 and -I1003 with Pol3-C1009. Pol31-D304 and -K358 residues are highlighted as magenta spheres. Pol31-L324, -Y327 and -F328 residues are shown as red spheres. Note that Pol31-L324 is buried in our model and not available for intermolecular interactions. (B) Yeast two-hybrid assays were performed using pBTM116 plasmids carrying WT and mutated Pol31 fused to the lexA binding domain (BD) and pACT2 plasmids carrying WT or mutated Pol3 C-terminal amino acids 1032–1097 fused to the Gal4 activating domain (AD). Each spots illustrates an individual co-transformant obtained following transformation of CTY10-5d. β-galactosidase activity was tested using an overlay plate assay at 22°C and 30°C. (C) Survival curves obtained after UV-irradiation of Pol δ mutant strains in which the interaction between Pol3 CTD and Pol31 is impaired. (D and E) Survival curves obtained after UV-irradiation of PCNA ubiquitination-deficient Pol δ mutant strains. (D) The pol3-R1043G and pol31-D304N alleles do not suppress rad18Δ-associated UV sensitivity. (E) The pol31K358I allele does not affect the suppression of rad18Δ-associated UV sensitivity by pol3-ct (red arrow).

Fig 6. Physical interactions between PCNA and the proposed PIP-boxes of S. cerevisiae Pol32, Pol31 and Pol3.

The thermograms (upper panels) and binding isotherms (lower panels) of the calorimetric titrations of PCNA by the assayed peptides at 303 K are presented. The corresponding thermodynamic parameters are reported in Table 1. (A, B, C, D) Interaction between PCNA and the peptides containing the (putative and canonical) Pol3, Pol31, Pol32 and Msh6 PIP motifs, respectively. The thermodynamic parameters ΔH, N, and Ka were obtained by non-linear least-squares fitting (represented in line mode) of the experimental data with a single site model.

Fig 7. Suppression of rad18Δ UV phenotypes by pol3-FI.

(A) Survival curves of UV-irradiated mutant strains carrying the pol3-FI, pol31-YF or pol32-pip allele. (B) pol3-FI suppresses rad18Δ-associated UV sensitivity (red arrow), while pol31-YF and pol32-pip do not. (C) Suppression of rad18Δ-associated UV sensitivity by pol3-FI (red arrow) depends on REV3. (D) Increase of UV-induced mutagenesis in rad18Δ cells carrying pol3-FI. Each dot represents the can^R frequency measured in one experiment. The median for each strain is represented by a horizontal bar: **, P = 0.0026 (Mann-Whitney test).

Fig 8. Polymerase switching in Pol δ structural mutants: A working model.

(A) DNA synthesis on an undamaged DNA template is catalyzed by Pol δ. Thanks to PCNA trimeric structure, simultaneous binding of two or even three Pol δ molecules could be envisaged. This possibility is not depicted here for the sake of simplicity. Pol δ outcompetes TLS polymerases for the access to the 3’ DNA primer terminus by stronger affinity or mass action. Pol δ subunits interact with several PCNA domains. The PCNA domain interacting with PIP motifs is represented by an orange rectangle on PCNA, whereas the one interacting with the CysA module of Pol3-CTD and possibly with the CysA module of Rev3-CTD is represented by a blue filled circle (note that this PCNA domain has not been identified yet). (B) Pol δ stalls when the 3’ end of the newly synthesized strand encounters a UV-induced DNA lesion such as a pyrimidine dimer. Rad6/Rad18 activation triggers PCNA ubiquitination at K164 (red filled circle). TLS polymerases access to the 3’ end is envisaged in three steps: 1) the TLS polymerase-ubiquitin interaction brings these polymerases closer to their PCNA interacting domain and therefore increases their local concentration near the 3’ end; 2) thanks to this proximity, TLS polymerases can compete with Pol δ for binding to their specific PCNA interaction domain: PIP interacting domain (orange rectangle) for Pol η and a not-yet-identified PCNA domain (blue filled circle) for Pol ζ via its CysA module. Following its displacement, Pol δ disengages from binding to the primer terminus; 3) tight binding of TLS polymerases on the front face of PCNA allows the insertion of the 3’ extending end and the template UV-induced DNA lesion into the catalytic site of TLS polymerases for translesion synthesis. (C) Pol δ-ct (and Pol3-FI) leads locally to the destabilization of the Pol3-CTD-PCNA interaction. With the help of Rev1, this could facilitate Pol ζ competition for the common PCNA interacting domain. Thus, the initial binding step to ubiquitinated PCNA is bypassed and Pol ζ has access to the 3’ end in two steps only: 1) Pol ζ binds to its PCNA interacting domain (blue filled circle); and then 2) Translesion synthesis takes place. Conversely, Pol η does not have access to its PCNA interacting domain in pol3-ct. (D) When Pol δ carries the mutant Pol32-ΔPIP subunit, Pol η PIP domain should readily interact with PCNA PIP interacting domain (orange rectangle, step 1) and the requirement of PCNA ubiquitination should be bypassed. This hypothesis was proved to be false because the pol32-pip allele does not suppress UV sensitivity associated with rad18Δ. Thus, Pol η cannot take over DNA synthesis at UV-induced DNA lesions (step 2 indicated by a dashed arrowed line and a question mark). Pol η might require additional PCNA interactions to accommodate pyrimidine dimers on its catalytic site.