The inner side of yeast PCNA contributes to genome stability by mediating interactions with Rad18 and the replicative DNA polymerase δ

PCNA is a central orchestrator of cellular processes linked to DNA metabolism. It is a binding platform for a plethora of proteins and coordinates and regulates the activity of several pathways. The outer side of PCNA comprises most of the known interacting and regulatory surfaces, whereas the residues at the inner side constitute the sliding surface facing the DNA double helix. Here, by investigating the L154A mutation found at the inner side, we show that the inner surface mediates protein interactions essential for genome stability. It forms part of the binding site of Rad18, a key regulator of DNA damage tolerance, and is required for PCNA sumoylation which prevents unscheduled recombination during replication. In addition, the L154 residue is necessary for stable complex formation between PCNA and the replicative DNA polymerase δ. Hence, its absence increases the mutation burden of yeast cells due to faulty replication. In summary, the essential role of the L154 of PCNA in guarding and maintaining stable replication and promoting DNA damage tolerance reveals a new connection between these processes and assigns a new coordinating function to the central channel of PCNA.


Results
The pol30-L154A yeast strain is slow-growing and exhibits a DNA damage sensitive phenotype. We reported in a previous study the characterization of PCNA variants that we generated during a directed mutational screening of yeast POL30 32 . As part of that screen, we engineered a mutant in which the leucine at position 154 was replaced by alanine (L154A PCNA). The L154 amino acid is situated at the inner surface of the PCNA ring, at the end of an alpha helix (Fig. 1a). This position is not conserved evolutionally at the amino acid level, but the corresponding leucine, isoleucine, and phenylalanine residues found in different species all have hydrophobic sidechains (Fig. 1b). Because the hydrophobic sidechain points away from the center of the PCNA ring and forms part of a hydrophobic cavity (Fig. 1a), we surmised that the mutation would not affect the sliding of PCNA on DNA. Still, the pol30-L154A strain expressing the mutant PCNA from a yeast centromeric plasmid as a sole source of PCNA exhibited a discernible growth defect compared to the wild type (Fig. 1c). It raised the possibility, that due to the vicinity of L154 to the subunit interface, the mutation might affect trimer stability, thereby interfering with replication. Indeed, examination of the wild-type and mutant PCNA by non-denaturing polyacrylamide gel electrophoresis exposed the monomer nature of the mutant, whereas the wild type existed as a trimer in vitro (Fig. 1d). Additionally, further investigation revealed that the mutation also increased the sensitivity of the pol30-L154A strain to DNA damaging agents compared with the wild type (Fig. 1e). These results suggested that whereas the mutation influenced replication, it also compromised the activity of one or more DNA damage response (DDR) pathways.
The L154A mutation in PCNA inactivates the RAD6/RAD18-dependent DNA damage tolerance. First, we investigated the possible link between DDR and the PCNA mutation. Applying epistasis analysis, we found that neither nucleotide excision repair (NER) nor homologous recombination (HR) was compromised by the PCNA mutation, since deletion of RAD14 or RAD52, the central regulators of NER and HR, respectively, in the pol30-L154A strain resulted in highly increased sensitivities upon both UV and MMS treatment, compared to the parental strains ( Fig. 2a, b). Surprisingly, deleting RAD18 governing DNA damage tolerance did not change the sensitivity of the pol30-L154A mutant (Fig. 2c). In other words, the high sensitivity of rad18Δ was suppressed to the sensitivity of the pol30-L154A strain. Moreover, the extreme sensitivity of rad6Δ was partially suppressed by the pol30-L154A mutation (Fig. 2d). To verify that the suppression was due to the PCNA mutation, we replaced the mutant PCNA expressing plasmid with one expressing the wild-type PCNA by plasmid shuffling in rad18Δ. Indeed, the expression of wild-type PCNA restored the high sensitivity of rad18Δ, confirming that the L154A amino acid change in PCNA alleviated the defect resulting from the absence of Rad18 (Fig. 2c). Testing several genes from the different sub-pathways of the Rad6/Rad18-governed DDR revealed that none of those genes could influence the DNA damage sensitivity of pol30-L154A (Fig. 2e). The results were confirmed using strains harboring pol30-L154A integrated into the POL30 genomic locus. Based on these data, we concluded that the Rad6/Rad18-dependent DNA damage tolerance was not functional in the pol30-L154A strain.
DNA damage-induced PCNA ubiquitylation and sumoylation at K164 are absent in pol30-L154A. One well-characterized PCNA mutation is known from the literature that bestows a very similar phenotype. That is the pol30-K164R mutation that abolishes the lysine where DNA damage-induced PCNA monoubiquitylation by Rad16-Rad18 or sumoylation during replication by Siz1 and Ubc9 takes place 14 . The lack of both modifications results in a moderate sensitivity to DNA damaging agents since the Rad52dependent homologous recombination, released from the inhibition of sumoylated PCNA, compensates for the sensitivity conferred by the inactivity of the Rad18 pathway 15 . We found that the pol30-L154A strain exhibited the same sensitivity to UV as the pol30-K164R strain (Fig. 3a). Moreover, the double mutant pol30-L154A-K164R strain was as sensitive as the single mutants. These results suggested that even the L154A mutation prevented both sumoylation and ubiquitylation of PCNA at K164. To corroborate this idea, we probed the DNA damageinduced ubiquitylation of PCNA in cells treated with 0.02% MMS. Though PCNA sumoylation occurs during normal replication, it is enhanced in response to sub-lethal doses of MMS 14 . Therefore, we also probed PCNA www.nature.com/scientificreports/ sumoylation in cells exposed to 0.3% MMS. As Fig. 3b, c shows, neither posttranslational modifications could be detected in the pol30-L154A or the pol30-K164R strain as opposed to the wild type demonstrating that the L154 residue is needed for both sumoylation and DNA damage-induced ubiquitylation of PCNA at K164.
The L154 residue of PCNA mediates direct interaction with Rad18. To get ubiquitylated at K164, PCNA has to interact with the ubiquitin ligase Rad18. We investigated the possibility that the L154A mutation hindered the binding of Rad18 to PCNA using yeast two-hybrid assays. Indeed, the L154A PCNA did not show interaction with Rad18, contrary to the wild type (Fig. 4a). In a control experiment, the L154A PCNA could still interact with Rad5, a known interacting partner of PCNA, indicating that the L154A mutation did not compromise the overall interaction capabilities of PCNA 14 . The role of the L154 amino acid in binding the SUMO ligase Siz1 could not be examined in similar assays as we could not detect the Siz1-PCNA interaction by yeast two-hybrid, in accord with a previous report (data not shown) 36 . Whether the L154 residue was directly involved in Rad18-binding was assessed in in vivo cross-linking experiments based on the site-specific insertion of the UV-A-inducible cross-linker amino acid p-benzoyl-Lphenylalanine (pBPA) 37 . Due to the presence of phenylalanine at the corresponding position in flies and worms, we reasoned that the phenylalanine derivative pBPA would maintain the hydrophobicity of the surroundings   www.nature.com/scientificreports/ prepared from UV-A-treated or untreated cells was followed by a tag-specific antibody. As Fig. 4b shows, a Rad18-PCNA cross-linked product appeared in UV-A-irradiated samples indicating that the modified phenylalanine pBPA residue directly mediated the Rad18-PCNA interaction. This observation is in agreement with the results obtained by yeast two-hybrid and together they infer a direct role for the L154 of PCNA in binding Rad18. However, as the experiment was carried out in the absence of any DNA damaging treatment, we could not conclude whether Rad18 was bound to free PCNA or to PCNA loaded on DNA.
The L154A PCNA mutation affects replication. Several mutations had been identified at the PCNA subunit interface which caused trimer instability as judged by the monomeric behavior in vitro, but still supported normal cellular growth suggesting that in vivo the mutants formed a trimer during replication [38][39][40] . It applied that the trimer instability observed with the L154A PCNA in vitro was not necessarily the cause of the observed replication defect of the mutant strain. Therefore, we set out to further characterize the effect of pol30-L154A on replication. The pol30-L154A strain exhibited marked sensitivity to the replication inhibitor hydroxyl-urea (HU) (Fig. 3a). Because the pol30-K164R strain devoid of sumoylation and DNA damageinduced PCNA ubiquitylation was as sensitive as the wild type, and the double mutant pol30-L154A-K164R strain showed the same HU-sensitivity as the pol30-L154A single mutant, we concluded that the influence on replication was independent of the posttranslational modifications of PCNA at K164. To further characterize its replication deficiency, we measured mutagenesis in the mutant strain. As expected from the inactivity of the RAD18 pathway, induced mutagenesis was abolished in the pol30-L154A strain (Fig. 5a). Importantly, though, spontaneous mutagenesis was ~ 5-fold higher in the pol30-L154A strain as compared to the wild type ( Fig. 5b) 21 .
To examine the involvement of TLS polymerases, we tested the effects of the single gene deletions of REV3, REV1, and RAD30 in the pol30-L154A strain. However, none of the deletions affected the rate of spontaneous www.nature.com/scientificreports/ mutations, and even the pol30-L154A rev3Δ rev1Δ rad30Δ quadruple mutant exhibited the same high mutation rate as the pol30-L154A strain. These observations demonstrated that the elevated spontaneous mutation rate in the pol30-L154A strain was independent of the TLS DNA polymerases.
Polymerase slippage frequently occurs in the pol30-L154A strain. Our results suggested that the replicative polymerase was responsible for generating excess spontaneous mutations in the pol30-L154A strain. Therefore, we investigated which activity of Polδ was affected by the L154A PCNA by generating genomic mutations in the Polδ exonuclease and polymerase domains, separately, and comparing the rate of spontaneous mutagenesis of the single mutants to double-mutant strains carrying the pol30-L154A mutation, additionally. As Fig. 6a shows, the D321A E323A amino acid changes in the Pol3 exonuclease active site (pol3-01) increased the rate of spontaneous mutations ~ 20-fold compared to the wild type 41,42 . However, in the presence of the pol30-L154A allele, the mutation rate was further elevated to ~ 50-fold. On the other hand, the D643N mutation in the polymerase domain (pol3-t) caused a ~ 5-fold increase in the mutation rate, which remained at the same level even in the pol30-L154A pol3-D643N double mutant strain 42,43 . These data showing a synergistic relationship between pol30-L154A and pol3-DE321,323AA, and epistasis between pol30-L154A and pol3-D643N revealed, that the observed increase in spontaneous mutagenesis was due to replication errors linked to the polymerase activity of Polδ, whereas the exonuclease activity of the enzyme was not affected by the PCNA mutant. The pol3-D643N mutation induces frameshifts in homonucleotide runs and repeat-associated extended deletions due to replication slippage 42 . To identify the mutations in the pol30-L154A strain, we sequenced the whole CAN1 gene from mutant colonies grown on canavanine-containing plates. As Fig. 6b shows, the most prevalent change detected in pol30-L154A was single-nucleotide deletion at short homonucleotide T or A runs accounting for more than 40% of the mutations, in contrast to the ~ 10% observed with the wild type. Additionally, ~ 15% of the mutations were single-nucleotide insertions in the mutant as opposed to the ~ 3% in the wild type. Meanwhile, C to G mutations, generally considered to be produced by the TLS DNA polymerase Rev1, dropped from ~ 15% found in the wild type to ~ 1% in the mutant. Together, these results suggested that misalignment events at the primer end provoked replication slippage in the pol30-L154A strain resulting in one nucleotide deletions/ insertions at short homonucleotide runs. The mutation spectrum of the Polδ mutant pol3-D643N exhibited a striking difference from that of pol30-L154A in that single nucleotide deletions accounted for only a few percent of the mutations, whereas extended deletions with a median length of ~ 80 nucleotides caused more than 40% of spontaneous mutations. Importantly, the double mutant pol3-D643N pol30-L154A behaved like pol30-L154A in that the ratio of single nucleotide deletions was as high as in pol30-L154A, whereas the ratio of extended deletions dropped back to a few percent of the wild-type level. It indicated that the pol30-L154A mutation was dominant over the pol3-D643N mutation.
The Polδ-PCNA complex is altered in the pol30-L154A strain. The above results suggested that the stability of the replication complex was extenuated in pol30-L154A. This prompted us to examine the interaction between PCNA and Polδ. Polδ consists of three peptides. Pol3 is the catalytic subunit, containing both the www.nature.com/scientificreports/ polymerase and exonuclease active sites. Pol31 is essential, whereas Pol32 is a non-essential accessory subunit 44 . All three subunits of Polδ have been shown to contain sequence motifs resembling the conserved PIP motif, and mutating any of them decreased DNA synthesis by Polδ in vitro in the presence of PCNA, RFC, and RPA 45 . Our experiments, testing the interactions of the three subunits of Polδ by yeast two-hybrid, revealed that Pol32 interacted with both the wild-type and the L154A PCNA, though the interaction with the mutant was weakened as evidenced by the lack of growth of the corresponding strain on 3-AT containing medium that provides more stringent selection for interactions (Fig. 6c). Moreover, Pol31 did not show interaction with the mutant PCNA, only with the wild type. In contrast, association with Rad5, Polη, and Apn2 were not affected 9,14,46 . We note, that the Pol3-PCNA interaction could not be examined in these assays as there was no detectable interaction even when using the wild-type PCNA construct. In summary, these results suggested that the L154A PCNA mutation weakened the interaction of the Polδ with PCNA.

Discussion
In this study, we have identified a PCNA residue that is essential for both DNA damage tolerance and normal replication through mediating interactions with key proteins of the two processes. The L154A PCNA mutation is situated at the inner side of PCNA, on the front of the molecule. It confers yeasts growth defect, HU sensitivity, and elevated spontaneous mutagenesis resulting from increased rates of single-nucleotide insertions and deletions at homopolymer nucleotide runs, all indicating difficulties during www.nature.com/scientificreports/ replication. The mutation causes trimer instability in vitro probably due to its proximity to the interdomain surface, which might lie behind the observed in vivo replication defects. However, this conclusion is contradicted by several PCNA mutations that render PCNA monomer in vitro, but do not lead to phenotypes associated with replication errors, when expressed in yeast cells. For example, the PCNA protein with the G178S amino acid change at the subunit interface is a monomer in vitro, but the corresponding yeast strain grows at the wildtype rate 39,40 . The same is true for the pol30-41 strain having the A112T and S135F replacements in PCNA 38 . In addition, pol30-41 exhibits a wild-type rate of spontaneous mutagenesis. These data indicate that the inability of PCNA to form trimers in vitro does not necessarily manifest in vivo. In any case, our experiments showing decreased binding of Pol31 and Pol32 to L154A PCNA suggest, that the replication defect of the mutant is caused by faulty interaction between PCNA and Polδ. Recent studies reported that only the catalytic subunit of Polδ interacts with PCNA and the interaction is mediated by the CysA element of Pol3 containing a Zn-finger and a PIP motif, whereas the regulatory subunits are positioned away from the PCNA ring [47][48][49] . The crystal structure of Polδ with PCNA and DNA also reveals that the amino acids involved in Pol3-binding are situated at the IDCL of PCNA, which is far from the L154 residue. In light of these, it is interesting that Pol32 could interact with the wild-type and though with decreased affinity also with the mutant PCNA in yeast two-hybrid, whereas Pol31 showed interaction only with wild-type PCNA. As the PIP motif at the C-terminus of Pol32 was shown to bind PCNA in isothermal calorimetric experiments and yeast two-hybrid, the interaction we observed is probably direct 50,51 . On the other hand, the interaction of Pol31 with PCNA could be indirect, mediated by endogenous Pol3. In either case, as in yeast two-hybrid PCNA can probably not form a trimer due to its fusion to the Gal4 activation domain, we can rule out that trimer instability affected the interactions. We surmise, that the L154 www.nature.com/scientificreports/ PCNA residue might directly bind Polδ and its role would be to help correct positioning of the polymerase during the establishment of the DNA-PCNA-Polδ complex. Hence, in its absence, an erratic complex is formed, in which Polδ is prone to dis-attach from the primer end. Interestingly, the L154 residue is also directly involved in Rad18 binding, as we show by yeast two-hybrid and direct in vivo cross-linking experiments. The failure of Rad18 to interact with L154A PCNA prevents DNA damage-induced PCNA ubiquitylation, evidenced by the lack of ubiquitylated PCNA upon MMS treatment. Accordingly, the Rad18-governed DDT is completely inactive in pol30-L154A. Nevertheless, the pol30-L154A strain is much more resistant to DNA damaging agents than the rad18 strain. Its equal sensitivity to and epistatic relationship with pol30-K164R, together with the lack of sumoylated PCNA in cells indicate that the L154A mutation prevents PCNA sumoylation at K164, as well, though we could not detect a direct Siz1-PCNA interaction using in vivo cross-linking. However, an important difference between pol30-L154A and pol30-K164R is that they exhibit differing UV sensitivities in rad52Δ background. The pol30-K164R rad52Δ double mutant is as sensitive as the rad6Δ rad52Δ strain indicating that HR released from the suppression of sumoylated PCNA compensates for the inactivity of the RAD6 pathway in pol30-K164R 15 . In contrast, we have found, that the pol30-L154A rad52Δ strain is less sensitive than the rad18Δ rad52Δ mutant suggesting that a DNA damage tolerance mechanism is activated in pol30-L154A rad52Δ cells.
We note, that the K196V mutation of yeast PCNA was described as defective in the interaction with Rad18, and it bestowed the same sensitivity to yeasts as the K164R PCNA mutation 52 . Based on this, we infer that sumoylation at K164 is also hindered in the pol30-K196V mutant. This, together with our findings suggest that Rad18 and a sumoylation enzyme, probably Siz1, connect PCNA through the same residues, this way ensuring the mutually exclusive ubiquitylation and sumoylation at K164. Curiously, closer examination of the positions of the K196, L154, and K164 residues in the crystal structure of PCNA revealed that they could be connected through a circle around the PCNA subunit. Using this information and also taking into consideration previous results showing that Rad18 works as a dimer which binds one Rad6 molecule, but the dimerization does not involve the Zn-finger of Rad18 and that the PCNA interacting surface was mapped to almost all parts of human Rad18, we set up a model of the PCNA-Rad18 interaction (Fig. 7) [53][54][55] . In this model, two Rad18 molecules form a circle around PCNA by binding each other through the RING domains at the inner side of the PCNA ring, whereas the Rad6-binding sites are at the outer surface of PCNA positioning Rad6 close to K164.
In summary, we surmise that the L154 PCNA residue forms part of a hydrophobic pocket mediating a direct interaction with Rad18 and probably with Polδ, which is abrogated by the lack of the hydrophobic side chain caused by the L154A mutation. We also conclude that the composite influence of the L154 amino acid of PCNA on genome stability foreshadows a regulatory mechanism that harmonizes replication and DNA damage response and demonstrates an emerging complexity of interactions mediated by the PCNA inner surface.
For quantification of UV sensitivity, cells were spread onto YPD plates at appropriate dilutions and irradiated with UV for given times to apply the required dosages. Colonies were counted after incubating the plates in the dark at 30 ℃ for 3-5 days.

UV-induced and spontaneous mutation rates. UV irradiation-induced forward mutation frequencies
at the CAN1 locus were measured by comparing the numbers of can1 R colonies at different UV doses, selected on synthetic complete (SC) -arg plates containing canavanine, with the number of colonies on SC plates exposed to the same UV doses. Spontaneous forward mutation frequencies were measured at the CAN1 locus. Ten single colonies for each strain were diluted in water. Approximately 10 cells from each colony were inoculated into 500 µl SC in a microtiter plate and grown for 3 days at 25 °C that is the permissive temperature for the pol3-t mutants. The cultures were plated onto canavanine containing SC-arg plates, and the average number of mutants in the case of each strain was counted for 10 7 plated cells. Mutation frequencies were determined using a chart based on the Lea-Coulson fluctuation model.
Sequencing of the CAN1 gene. For sequencing of the CAN1 gene, 3 canavanine resistant single colonies were picked from each plate of the spontaneous mutagenesis experiment and grown in 2 ml YPD overnight. Altogether, 84-84 colonies from the wild-type and the mutant strains were examined. Genomic DNA was isolated from the cultures, and the can1 gene was amplified with 5'-GAA GAG TGG TTG CGA ACA GAG-3' and 5'-CTT ATG AGG GTG AGA ATG CG-3' oligonucleotides. The PCR products were sequenced using the 5'-CTG TCA CGC AGT CCTTG-3' , 5'-GGT GAG AAT GCG AAA TGG -3' and 5'-ATA CTA ATC CAT GCC GCC -3' gene specific primers.
Growth assays. To compare the growth of the strains, a Synergy 2 automated microplate reader machine (BioTek, USA) was used. Aliquots of 1 μl of overnight starter cultures of the wild-type and the mutant pol30 expressing strains were transferred into 100 μl of fresh YPD in dedicated 96-well plates. Absorbance at 600 nm was measured every 20 min for 24 h at 30 °C with continuous shaking. Averages of 3 experiments with 3-3 technical parallels in each are shown.
Yeast two-hybrid. The binding and activation domain fusion constructs were co-transformed into PJ69-4A 58 , and transformants were selected on SC-leu/trp medium. Single colonies of the transformants were resuspended in water at equal densities and spotted onto an SC-leu/trp plates to demonstrate growth and onto an SC-leu/trp/his plates to assess the HIS3 gene expression reflecting protein-protein interaction. 3-amino-triazole (3-AT) was added to SC-leu/trp/his plates at defined concentrations to test the strength of the interactions. Plates were incubated at 30 ℃ for 3-5 days.
In vivo PCNA ubiquitylation and sumoylation. PCNA ubiquitylation was detected by applying denaturing Ni-nitriloacetic acid (Ni-NTA) chromatography 60 . Briefly, 200 ml yeast cultures grown in YPD to A 600 :0.8 were synchronized by alpha-factor for 3.5 h. After microscopic verification of the G1 state of the cultures, they were released back to cycling in fresh YPD, and half of the cultures were treated with 0.02% MMS for 90 min. After harvesting, cell pellets were frozen in liquid nitrogen. Following cell lysis in cold 1.85 N NaOH, 7.5% β-mercaptoethanol, proteins were precipitated by trichloroacetic acid (TCA) and resuspended in Buffer A (6 M guanidine HCl, 100 mM sodium phosphate, 10 mM Tris/HCl, pH 8.0) by rotating for 1 h at room temperature. Lysates were clarified and adjusted to pH ~ 8 then samples were bound to Ni-NTA agarose in the presence of 0.05% Tween-20 and 10 mM imidazole overnight. After binding, beads were washed with Buffer A and Buffer B (8 M urea, 100 mM sodium phosphate, 10 mM Tris, pH 6.3, 0.05% Tween-20), and bound proteins were eluted for 10 min at 60 °C in 20 µl of buffer C (8 M urea, 200 mM Tris/HCl, pH:6.8, 1 mM EDTA, 5% SDS, 0.1% bromophenol blue, and 1.5% DTT), and proteins were resolved on a 12% SDS-PAGE. Following blotting, the www.nature.com/scientificreports/ PVDF membrane (Immobilon-P, Merck Millipore Ltd.) was pretreated in 6 M guanidyl-HCl, 20 mM Tris/HCl pH:7.5, 5 mM 2-ME for 30 min at 4 °C to probe with anti-Ub antibody (2000×, Santa Cruz sc8017). An aliquot of the same samples was run on a second gel that was developed by a PCNA antibody. Anti-mouse (30,000×, Thermo Scientific 31430) and anti-rabbit (10,000×, Thermo Scientific 31460) secondary antibodies were applied. PCNA sumoylation was detected in whole-cell extracts. 15 ml of fresh cultures were grown in YPD to OD 0.7-0.8, synchronized by alpha-factor for 2 h, and treated with 0.3% MMS. At given time points 2 ml aliquots were collected, and whole-cell extracts were prepared using NaOH/TCA 60 . Pellets were resuspended in buffer C, and after boiling, proteins were resolved on a 12% SDS-PAGE. PCNA was visualized using anti-PCNA primary (20,000×) and anti-rabbit (10,000×, Thermo Scientific 31460) secondary antibodies.
In vivo UV-A induced protein cross-linking. 3 ml yeast cultures were incubated overnight in SC-ura/ trp to select for the PCNA-6His and the artificial amino-acil synthetase expressing constructs. The next day cultures were diluted in 50 ml fresh media containing 1 mM pBPA (p-benzoyl-L-phenylalanine), and further incubated overnight. Cultures were diluted to OD 600 : 0.15 in 50 ml fresh media. After 1.5 h of incubation, cells were collected, washed, and resuspended in 10 ml PBS. Half of the cultures were irradiated with UV-A (365 nm) for 30 min on ice in a Petri dish with constant shaking. Cells were broken as described earlier 61 . Proteins were eluted from the Ni-beads by boiling and resolved on an 8 and a 10% SDS-PAGE. Rad18 cross-linked to PCNA was visualized using anti-HA (10,000×, Abcam ab9110) and anti-rabbit (10,000×, Thermo Scientific 31460) antibodies. To detect PCNA the membrane was developed with anti-PCNA (5000×) and anti-rabbit (10,000×, Thermo Scientific 31460) antibodies. www.nature.com/scientificreports/