PCNA-Polκ-Polδ/USP18 axes stabilize replication fork and restart to reduce cisplatin cytotoxicity

Reviewer #1 (Public review):

Summary:

Cisplatin, a platinum-based chemotherapeutic agent, induces intra- and interstrand crosslinks, thereby blocking DNA replication and transcription and triggering apoptosis. The authors aim to demonstrate that DNA polymerase κ (Polκ), traditionally seen as a translesion synthesis (TLS) polymerase, able to synthesize DNA through DNA lesions, plays a non-catalytic, structural role in stabilizing replication forks and protecting cells from cisplatin-induced cytotoxicity. A key finding of this work is the identification of two novel molecular axes: PCNA-Polκ-Polδ, which facilitates efficient DNA replication; PCNA-Polκ-USP18, which stabilizes DNA damage response proteins. These findings provide actionable therapeutic targets for overcoming head and neck squamous cell carcinoma chemoresistance, a cancer with rising incidence and limited treatment options.

Strengths:

The study relies on a robust experimental design, including Polk allegedly CRISPR-Cas9 knockout, siRNA knockdown, and rescue experiments with wild-type, catalytically dead, and PCNA-interaction-deficient Polκ variants, supporting a non-catalytic role of Polκ. The work also reports a strong implication of Polk in cisplatin resistance, the identification of USP18 as a possible Polk partner and the consequences of Polk depletion on post-translational stabilisation of DNA damage response proteins.

Weaknesses:

The findings reported in this manuscript cannot be generalized to all cisplatin resistance mechanisms, as cells may develop multiple adaptive strategies to survive chemotherapy. Polκ's role varies across cancer types. For example, it is downregulated in stomach and colorectal cancers but upregulated in HNSCC, lung, and ovarian cancers. Thus, its use as a biomarker or drug target may be context-dependent.

Acute cisplatin exposure is sufficient to trigger Polκ upregulation to levels similar to those in resistant cells. However, it remains unclear how long this upregulation persists and to what extent it contributes to survival. Further, the sensitivity of cisplatin-naïve H357 or SCC9 cells (H357-S and SCC9-S) to Polκ knockdown has not been addressed. This is a critical question, as acute cisplatin exposure induces Polκ expression to levels similar to those in resistant cells. This could argue against a direct role for Polκ in mediating resistance and instead suggest indirect mechanisms (like Polκ-dependent mutations during adaptation).

The experimental design and results aimed at demonstrating the existence of a PCNA-Polκ-USP18 axis (Figure 9A) do not fully support the conclusion that these proteins form a stable complex. This set of experiments also lacks essential controls, such as the immunoprecipitated bait and the amount of immunoglobulins precipitated in all conditions. This also applies to the colocalization experiments in cells shown in Figure 9B. Images are poor and lack quantification. Further, Polk is seen mainly cytoplasmic in the upper panel, while it is nuclear in the lower panel. Discrepancies in Polk subcellular localization are also evident in the Supplementary data. USP18 is known to deubiquitinate ISG15-modified proteins (not just ubiquitin). The study does not rule out ISGylation as a contributing mechanism. The experimental design involving analysis of DNA synthesis dynamics at a single-molecule level is not appropriate. Overinterpretation of the data in several parts of the manuscript and lack of rigor in performing the experiments. Inappropriate consideration and absence of discussion of previously published literature directly related to the subject studied in this manuscript. Discrepancy with a previous report regarding the role of Polk in Chk1 phosphorylation (Tonzi et al., eLife 2018). Synergic effect of T2AA inhibitor and Cisplatin have been already described in « naive » cancer cells (Inoue et al, 2014). Another critical point is that the proliferation rate of Polk-depleted cells is slower than that of wild-type cells. Hence, the colony formation assay shown in Figure 2B can be misleading, since the observed differences can be interpreted only as a proliferation problem.

Reviewer #2 (Public review):

Summary:

Building on earlier studies, the authors report a role for pol kappa in mediated cisplatin resistance. Their data on dispensability of pol kappa catalytic activity for cisplatin resistance is consistent with previous reports. They further demonstrate that the PIP box of pol kappa is critical for cisplatin response. Based on these observations, the study concludes that targeting pol kappa and PCNA interaction can be a viable approach to overcome cisplatin resistance.

Strengths:

Indications that interaction between Pol kappa PIP box and PCNA can be targeted to overcome cisplatin resistance.

Weaknesses:

(1) The study has used a model of cisplatin resistance and found that the phenotype is specifically reliant on upregulation of Pol kappa. They also observe that in this model of cisplatin resistance, there is rapid degradation of multiple repair proteins, including ATM, ATR, HR and NHEJ proteins upon knocking out Pol kappa. However, it is unclear how the resistant model was derived. Also, since the data and almost all experiments in this manuscript were performed with a single model of cisplatin resistance, the conclusions should be taken with caution.

(2) There are also inconsistencies in findings. Increased G2 arrest and no change in origin firing are being observed despite a significant reduction in Chk1 protein levels.

Reviewer #3 (Public review):

This manuscript investigates the role of PolK in cisplatin repair. While in general it is considered that polK is not involved in the repair of cisplatin-induced DNA damage, the authors show that in a very specific scenario, namely cisplatin-resistant head and neck cancer cells, loss of PolK causes cisplatin sensitization, implying a role in cisplatin repair by polK in these cells. It is also implied that these cells acquire cisplatin resistance by overexpressing polK, but this is not really investigated. The authors then go on to show that DNA replication in the presence of cisplatin is affected by the loss of polK in these cells and also identify USP18 as a potential polK interactor in these cells with a similar phenotype. They claim that polK and USP18 form a pathway that allows cisplatin tolerance in these cisplatin-resistant head and neck cancer cells. The findings are interesting and useful to the field; however, the manuscript, in its current form, has several issues. Most importantly, the mechanism of USP18 has not been investigated. In addition, the manuscript does not flow fluidly, and instead, various experiments are put together without a clear logic. Some of the claims are not substantiated by the data shown.

(1) The experiments in Figure 1 using a few cell lines from various types of cancers are not enough to conclude that polK expression is specifically induced by cisplatin in some types of cancers but not others. Since the focus of this study is head and neck cancer, the authors should show the expression of PolK after cisplatin treatment in more head and neck cancer cell lines, and not just the two investigated.

(2) It is unclear to me why the authors include H357-S in their experiments. If the idea is that these cells acquire resistance because they overexpress polK, then the authors should investigate this by exogenously overexpressing PolK in H357-S cells and test if these cells are cisplatin resistant.

(3) In addition, the authors should create the polK knockout in H357-S cells as well and include it as a control in their experiments.

(4) Page 6, line 28: the comet assay does not measure DNA degradation, but rather DNA breaks.

(5) Figure 4B: How does the overexpression of PolK mutants compare to endogenous PolK expression? It is important to assess if this expression is similar or of much higher magnitude.

(6) Page 9, line 22: "For such a function, the catalytic domain of PolK becomes dispensable, whereas its interaction with PCNA is sufficient to drive efficient replication". I do not understand what data the authors used to make this claim. The interaction and colocalization studies should be performed with the PIP mutant. Similarly, this mutant should be used in the HU DNA fiber assays.

(7) It is unclear how USP18 acts. What are its substrates? Chk1/2, BRCA1, BRCA2? This needs to be investigated. The impact of PolK on this activity needs to be assessed as well (is PolK needed for USP18-mediated de-ubiquitination of these DSBR proteins?). As it stands, the manuscript does not address the mechanism of USP18 in DNA repair, which is billed as the main finding of the paper.

(8) Do PolK and USP18 interact directly? Experiments using recombinant proteins would be useful to address this.

Author response:

Reviewer #1 (Public review):

Summary:

Cisplatin, a platinum-based chemotherapeutic agent, induces intra- and interstrand crosslinks, thereby blocking DNA replication and transcription and triggering apoptosis. The authors aim to demonstrate that DNA polymerase κ (Polκ), traditionally seen as a translesion synthesis (TLS) polymerase, able to synthesize DNA through DNA lesions, plays a non-catalytic, structural role in stabilizing replication forks and protecting cells from cisplatin-induced cytotoxicity. A key finding of this work is the identification of two novel molecular axes: PCNA-Polκ-Polδ, which facilitates efficient DNA replication; PCNA-Polκ-USP18, which stabilizes DNA damage response proteins. These findings provide actionable therapeutic targets for overcoming head and neck squamous cell carcinoma chemoresistance, a cancer with rising incidence and limited treatment options.

Strengths:

The study relies on a robust experimental design, including Polk allegedly CRISPR-Cas9 knockout, siRNA knockdown, and rescue experiments with wild-type, catalytically dead, and PCNA interaction-deficient Polκ variants, supporting a non-catalytic role of Polκ. The work also reports a strong implication of Polk in cisplatin resistance, the identification of USP18 as a possible Polk partner and the consequences of Polk depletion on post-translational stabilisation of DNA damage response proteins.

Thank you so much for appreciating our efforts to demonstrate role of Polκ mediated axes in cisplatin resistance in head and neck cancer cells.

Weaknesses:

The findings reported in this manuscript cannot be generalized to all cisplatin resistance mechanisms, as cells may develop multiple adaptive strategies to survive chemotherapy. Polκ's role varies across cancer types. For example, it is downregulated in stomach and colorectal cancers but upregulated in HNSCC, lung, and ovarian cancers. Thus, its use as a biomarker or drug target may be context-dependent.

We completely agree with you, and the presented data only support Polκ's role in HNSCC as demonstrated in both acute cisplatin exposure as well as the cisplatin-resistant HNSCC models. Other cell and cancer types may adopt different strategies for cisplatin resistance.

Acute cisplatin exposure is sufficient to trigger Polκ upregulation to levels similar to those in resistant cells. However, it remains unclear how long this upregulation persists and to what extent it contributes to survival. Further, the sensitivity of cisplatin-naïve H357 or SCC9 cells (H357-S and SCC9-S) to Polκ knockdown has not been addressed. This is a critical question, as acute cisplatin exposure induces Polκ expression to levels similar to those in resistant cells. This could argue against a direct role for Polκ in mediating resistance and instead suggest indirect mechanisms (like Polκ-dependent mutations during adaptation).

Since H357-S and SCC9-S cells are highly sensitive to cisplatin, knocking down of Polκ unlikely will alter the phenotype, as other TLS DNA polymerases like Polκ and Polκ play critical role in such lesion bypass. Since no other DNA polymerase was upregulated in these cells upon cisplatin exposure and in the cisplatin-resistant cells, it was intriguing to demonstrate a direct role of Polκ in chemoresistance and that has been proven in this study. Since the catalytic activity of Polκ is not required to induce chemoresistant in these cells, we strongly believe that Polκ-dependent mutagenesis play minimal or no role in adapting cells to tolerate cisplatin. Nevertheless, we will knock down Polκ in these cells and determine cisplatin sensitivity

The experimental design and results aimed at demonstrating the existence of a PCNA-Polκ-USP18 axis (Figure 9A) do not fully support the conclusion that these proteins form a stable complex. This set of experiments also lacks essential controls, such as the immunoprecipitated bait and the amount of immunoglobulins precipitated in all conditions. This also applies to the colocalization experiments in cells shown in Figure 9B. Images are poor and lack quantification. Further, Polk is seen mainly cytoplasmic in the upper panel, while it is nuclear in the lower panel. Discrepancies in Polk subcellular localization are also evident in the Supplementary data.

We appreciate the Reviewer's critical and insightful comment. In our view, the interaction between Polκ and USP18 is very specific as USP2 and IgG alone do not pull down Polκ. Similarly, we also show that both Polκ and USP18 interact with PCNA. We agree with the reviewer that the existence of a stable complex of PCNA-Polκ-USP18 has not been fully demonstrated in the current version. We will perform additional experiments to strengthen our finding: a) Co-IP experiments with Polκ PIP mutants (wild-type vs. mutant) should be performed to determine whether USP18 loses its ability to bind PCNA in the absence of Polκ-PCNA interaction. b) Mapping the domain in Polκ that is involved in USP18 binding and their Co-IP experiment. Additionally, high resolution co-localisation images including quantified data will be provided.

USP18 is known to deubiquitinate ISG15-modified proteins (not just ubiquitin). The study does not rule out ISGylation as a contributing mechanism.

We find the point raised by the reviewer is very intriguing, however, as it will require a significant amount of time and effort to demonstrate ISGylation of DDR proteins and deISGylation by UPS18, and the insight that we may gain is unlikely to add to the central theme of this paper, we will expand this in our subsequent related study. Thank you for the suggestion.

The experimental design involving analysis of DNA synthesis dynamics at a single-molecule level is not appropriate. Over interpretation of the data in several parts of the manuscript and lack of rigor in performing the experiments. Inappropriate consideration and absence of discussion of previously published literature directly related to the subject studied in this manuscript. Discrepancy with a previous report regarding the role of Polκ in Chk1 phosphorylation (Tonzi et al., eLife 2018). Synergic effect of T2AA inhibitor and Cisplatin have been already described in « naive » cancer cells (Inoue et al, 2014).

Thank you very much for the suggestions. We will take care of the portions and modify as suggested. The necessary reference will be added as appropriate.

Another critical point is that the proliferation rate of Polk-depleted cells is slower than that of wild-type cells. Hence, the colony formation assay shown in Figure 2B can be misleading, since the observed differences can be interpreted only as a proliferation problem.

Thank you for pointing this out and we will modify the portion for better clarity.

Reviewer #2 (Public review):

Summary:

Building on earlier studies, the authors report a role for pol kappa in mediated cisplatin resistance. Their data on dispensability of pol kappa catalytic activity for cisplatin resistance is consistent with previous reports. They further demonstrate that the PIP box of pol kappa is critical for cisplatin response. Based on these observations, the study concludes that targeting pol kappa and PCNA interaction can be a viable approach to overcome cisplatin resistance.

Strengths:

Indications that interaction between Pol kappa PIP box and PCNA can be targeted to overcome cisplatin resistance.

Thank you for appreciating our finding that the PIP box of Polκ is critical for cisplatin response

Weaknesses:

(1) The study has used a model of cisplatin resistance and found that the phenotype is specifically reliant on upregulation of Pol kappa. They also observe that in this model of cisplatin resistance, there is rapid degradation of multiple repair proteins, including ATM, ATR, HR and NHEJ proteins upon knocking out Pol kappa. However, it is unclear how the resistant model was derived. Also, since the data and almost all experiments in this manuscript were performed with a single model of cisplatin resistance, the conclusions should be taken with caution.

We are extremely sorry for the lack of clarity. Please note that two cisplatin-resistant models (H357 and SSC9) have been used and the results were very consistent in both cells. Fig. 1C clearly demonstrates about the generation of these resistant models and the original reference has been already cited.

(2) There are also inconsistencies in findings. Increased G2 arrest and no change in origin firing are being observed despite a significant reduction in Chk1 protein levels.

Thank you for pointing this out. In our view, the increased G2 arrest is due to more fork stalling or collapsed than the new origin firing. Also, in our assay we observed less than 10% of new origin fired DNA fibres, and that could be the reason of no significant change in new origin firing among various cells.

Reviewer #3 (Public review):

This manuscript investigates the role of PolK in cisplatin repair. While in general it is considered that polK is not involved in the repair of cisplatin-induced DNA damage, the authors show that in a very specific scenario, namely cisplatin-resistant head and neck cancer cells, loss of PolK causes cisplatin sensitization, implying a role in cisplatin repair by polK in these cells. It is also implied that these cells acquire cisplatin resistance by overexpressing polK, but this is not really investigated. The authors then go on to show that DNA replication in the presence of cisplatin is affected by the loss of polK in these cells and also identify USP18 as a potential polK interactor in these cells with a similar phenotype. They claim that polK and USP18 form a pathway that allows cisplatin tolerance in these cisplatin-resistant head and neck cancer cells. The findings are interesting and useful to the field; however, the manuscript, in its current form, has several issues. Most importantly, the mechanism of USP18 has not been investigated. In addition, the manuscript does not flow fluidly, and instead, various experiments are put together without a clear logic. Some of the claims are not substantiated by the data shown.

Thank you very much for finding our study interesting and the pending concerns will be addressed as suggested.

(1) The experiments in Figure 1 using a few cell lines from various types of cancers are not enough to conclude that polK expression is specifically induced by cisplatin in some types of cancers but not others. Since the focus of this study is head and neck cancer, the authors should show the expression of PolK after cisplatin treatment in more head and neck cancer cell lines, and not just the two investigated.

In this study, we have explored eight different cell types (breast, brain, liver, head and neck, pancreatic, prostrate, lungs, and kidney) to check the expression of Polκ upon cisplatin exposure, and HNSCC cells only showed Polκ up-regulation. Therefore, we went ahead for further demonstration of the role of Polκ in cisplatin resistance in OSCC using four different cell models (H357-S, H357-R, SSC9-S, and SSC9-R). By adding more cell lines to study will unlikely change the central theme of the paper. Yes, by acquiring and analysing clinical samples from the cisplatin responder and non-responders would have strengthen our finding.

(2) It is unclear to me why the authors include H357-S in their experiments. If the idea is that these cells acquire resistance because they overexpress polK, then the authors should investigate this by exogenously overexpressing PolK in H357-S cells and test if these cells are cisplatin resistant.

It’s an interesting point and we will check whether overexpression of Polκ in H357-S cells could induce resistance to cisplatin and alters IC₅₀. Thank you for the suggestion.

(3) In addition, the authors should create the polK knockout in H357-S cells as well and include it as a control in their experiments.

We appreciate your suggestion. As suggested by Reviewer #1 also, we will check the phenotype of Polκ knockdown H357-S cells.

(4) Page 6, line 28: the comet assay does not measure DNA degradation, but rather DNA breaks.

Thank you for the suggestion, we will modify the text accordingly.

(5) Figure 4B: How does the overexpression of PolK mutants compare to endogenous PolK expression? It is important to assess if this expression is similar or of much higher magnitude.

Please note that GFP-Polκ has been overexpressed in H357 Polκ knockout cells to nullify the effect of endogenous Polκ, otherwise we will not be able to test the role of various Polκ mutants.

(6) Page 9, line 22: "For such a function, the catalytic domain of PolK becomes dispensable, whereas its interaction with PCNA is sufficient to drive efficient replication". I do not understand what data the authors used to make this claim. The interaction and colocalization studies should be performed with the PIP mutant. Similarly, this mutant should be used in the HU DNA fiber assays.

We are extremely sorry for the lack of clarity. The inference has been derived from two sets of experiments as shown in Fig. 4C and Fig. 4D (and is with HU).

(7) It is unclear how USP18 acts. What are its substrates? Chk1/2, BRCA1, BRCA2? This needs to be investigated. The impact of PolK on this activity needs to be assessed as well (is PolK needed for USP18-mediated de-ubiquitination of these DSBR proteins?). As it stands, the manuscript does not address the mechanism of USP18 in DNA repair, which is billed as the main finding of the paper.

It has already been demonstrated in Fig. 9C where by knocking down USP18, the DDR proteins like Chk1, Chk2, CtIP, and Artemis can be recovered for ubiquitin-mediated proteasomal degradation. The same results are also obtained when its interacting partner Polκ is deleted. In our view, the presented results have sufficiently demonstrated the role of Polκ-Usp18 in the repair of cisplatin adducts through DDR proteins.

(8) Do PolK and USP18 interact directly? Experiments using recombinant proteins would be useful to address this.

We appreciate your suggestion. Since the Usp18 protein is not readily available, we will not be able to show; however, we believe the interaction is direct, and we will be able to map the binding site in Polκ.