Pre-activation of autophagy impacts response to olaparib in prostate cancer cells

Poly (ADP-ribose) polymerase 1 (PARP1) plays an essential role in DNA repair and is targeted by anticancer therapies using PARP inhibitors (PARPi) such as olaparib. PARPi treatment in prostate cancer (PC) is currently used as a monotherapy or in combination with standard therapies (hormonotherapy) in clinical trials for patients with DNA damage response mutation. Unfortunately, 20% of these patients did not respond to this new treatment. This resistance mechanism in PC is still not well understood. Here, we report that autophagy affects differently the response of PC cell lines to olaparib depending on its activation status. Pre-activation of autophagy before olaparib resulted in an increase of DNA repair activity by homologous recombination (HR) to repair double-strand breaks induced by olaparib and enhanced cell proliferation. When autophagy was activated after olaparib treatment, or completely inhibited, PC cells demonstrated an increased sensitivity to this PARPi. This autophagy-mediated resistance is, in part, regulated by the nuclear localization of sequestrosome 1 (SQSTM1/p62). Decrease of SQSTM1/p62 nuclear localization due to autophagy pre-activation leads to an increase of filamin A (FLNA) protein expression and BRCA1/Rad51 recruitment involved in the HR pathway. Our results reveal that autophagy basal levels may in part determine amenability to PARPi treatment.

There are several major concerns about the rest of the data. Unfortunately, the rest of the data is not consistent between the cell lines nor with repeat experiments shown in supplementary data. The authors claim that in Fig 2B and S1B, rapamycin pretreatment, but not posttreatment leads to increased LC3II and decreased p62 in all lines. This is seen in LNCaP and C4-2B in Fig 2B, but only for LNCaP in Fig S1B and never for PC3. Furthermore, rapamycin or olaparib alone is often having the same impact as the combined treatment. Same applies to the cell cycle analysis (Fig. 2D, S1C), where C4-2B seems to act as predicted, but the other lines are inconsistent. Data in figures 3 and 5 don't include the post-treatment data to be able to directly compare to the pre-treatment data within the same experiment. Instead the post-treatment data is placed with another set of data in supplementary figures 2,3, and 4. The quantification between these different sets of data leads to differences in results and differences in statistical significance. Interpretation was also complicated by the fact that Supplementary Fig 3 data seems to be superimposed over Supplementary Fig 2 data. This questions whether the stats presented are simply technical replicates within each experiment and not biological replicates. Figure 5C would be strengthened by demonstrating that p62 is also nuclear localized in these extracts. It is not clear why the quantification of the blots in Fig5C were normalized to olaparib rather than to untreated controls. The effect on filamin A is levels is modest at best. Quantification of several different blots from different experiments is required to assure statistical significance and consistency.

Data in
Overall these results are only correlative with respect to Rad51/Brca2, p62, and filamin A and their presumed dependency/relationship. There are no experiments that disrupt expression/activity of any of these proteins and show that they impact the response to olaparib. Authors need to be careful about using strong statements indicating their data support their proposed model.
The authors should also acknowledge that blocking ATG16L1 or autophagy in general will result in p62 accumulation simply because you are blocking its degradation.
It should be noted that C4-2B and LNCaP display a dramatic response to olaparib (log-fold) upon autophagy inhibition, but PC-3 displays less than a 2-fold increase in sensitivity, indicating at least for PC-3 cells, there are other factors involved. The authors need to acknowledge and address this more in the Discussion.
Reviewer #3 (Remarks to the Author): Cahuzac et al investigated the impact of autophagy on prostate cancer cell response to the PARPi olaparib by following the autophagy activation timeline. And Cahuzac et al define that preactivation of autophagy before olaparib treatment reduces the level of SQSTM1/p62 in the nucleus, results in an increase of DNA repair activity by homologous recombination to repair double-strand breaks induced by olaparib, and enhances cell proliferation. The authors should design and perform more experiments to support their conclusions.
indicating similar levels of autophagy activation" (Fig. 2B). The cell proliferation was significantly increased under RO10 conditions, but not under O10R conditions (Fig. 2C). Even under O10R conditions, the proliferation of LNCaP, C4-2B and PC-3 cells was decreased compared to olaparib treatment alone (Fig. 2C). The authors should perform another experiment (may choose cell viability assay using CellTiter-Glo) to further confirm the different anti-proliferation effect under RO10 and O10R conditions, which both showed a similarly high level of autophagy activation. 4. In Fig.2B, the western blots showed that in C4-2B WT cells, the autophagy was not activated, even was inhibited in all the groups with Rapamycin treatment, compared to the Control group as shown by LC3-II and p62. The authors should further confirm that. For these western blots, the authors may also show the normalized values or plots explaining the change in expression normalized to Actin. 5. In Fig. 2C, it showed that the olaparib treatment (O10) inhibited the proliferation of PC-3 Atg16L1 KO cells more than that of PC-3 WT cells. But in Fig. 1G-H, the authors have demonstrated that complete depletion of autophagy reduced the olaparib IC50 value from2.13 μM to 1.5 μM for PC-3. The authors should further confirm that.
6. The authors showed that pre-activation of autophagy limited the effect of olaparib on cell proliferation and cell cycle and this protective effect was abrogated when autophagy was abrogated by Atg16L1 KO (Fig.2). The authors should do one more knockout rescue experiment to see if the re-activation of autophagy would make the PC cells sensitive to olaparib again. 7. The authors showed that using their plasmid-based DNA repair reporter assays, complete depletion of autophagy by Aty16L1 KO significantly reduced the efficiency of HR (Fig.4F). Could the efficiency of HR be rescued with re-expression of Aty16L1? 8. The authors mentioned "we observed that the KO cell lines had higher expression levels of SQSTM1/p62 compared to WT cells (

Reviewer #1 (Remarks to the Author):
Comments to the authors: The Authors demonstrated that autophagy plays a role in the cellular response to olaparib treatment in pancreatic cancer cell lines. Such role might be mediated by the regulation of DNA repair efficiency, in particular through the inhibition of SQSTM1/p62 nuclear localization and the subsequent increase of FLNA, which is directly involved in the recruitment of proteins belonging to the HR pathway. The manuscript is well written and provides potential translational applications, such as the evaluation of autophagy basal levels as a strategy to predict patient responsivity to PARPi therapy and the introduction of combination therapies with PARPi and autophagy inhibitors to overcome PARPi resistance.
However, the following concerns should be addressed: To address this point, we have modified the manuscript to include the appropriate reference as follows in line 169: In C4-2B cells, after 6 days of O10 treatment, 70% of cells were blocked in S phase compared to 7% of control cells as it was shown in different cell lines in the literature (Yang, Ndawula et al. 2015, Pirotte, Holzhauser et al. 2018, Mani, Jonnalagadda et al. 2019).

Moreover, the FACS analysis of cell cycle distribution reported a very low percentage of cells in S phase in the control, which is unusual for cancer cell lines. Please clarify these findings.
While there are instances where the distribution in S phase is high, it is not without precedence to see the proportion of cancer cells in S phase vary between 5-30% depending on the cell lines used and cellular confluence, especially in time course experiments (Bort, Quesada et al. 2018, Sun, Weng et al. 2018, Wu, Peng et al. 2021. Therefore, the percentages we observed (15% for LNCaP, 7% for C4-2B and 10% PC-3 at day 2). Since autophagy is compared between the experimental groups and not the untreated control, we feel confident in our conclusions.

Probably, to confirm the effect of Olaparib on cell cycle distribution, it would be useful to check by WB experiments the expression of proteins related to cell cycle checkpoints, such as Cyclin B1, Cdc25C, p21 and Cyclin D1
While we agree that the suggestion is interesting in itself, it would not impact the conclusions of the study. In addition, the relationship between these proteins and the cell cycle have been well documented and thus no additional experiments were performed.
4. The authors should add more details in the discussion, focusing on current therapies for PC and the impact of PARPi resistance on patient survival. This would help the reader to understand the significance of findings illustrated in this study.
This would indeed help the reader better understand the impact of our findings. We determined that the information could be introduced early on to better contextualize our work. Therefore, we added the following paragraph in the introduction to clarify this point between line 45 and 56. One avenue of intense research in this area is in the use of PARP inhibitors in prostate cancer.

Reviewer #2 (Remarks to the Author):
In this manuscript, the authors use LNCaP, C4-2B, and PC-3 prostate cancer cells lines to determine how autophagy causes resistance to olaparib treatment. They use rapamycin to induce autophagy, and CRISPR KO of ATGp16L1 to block autophagy. Based on their results using these two manipulations, they propose a model whereby pre-activation of autophagy by rapamycin increases the sensitivity to olaparib by decreasing nuclear SQSTM1/p62, which increases homologous recombination-mediated repair through increased filamin A expression and recruitment of BRCA1/Rad51.
The data outlined in Figure 1 nicely demonstrates a strong correlation between high autophagy levels and increased resistance to olaparib is theses three cell lines, and that inhibiting autophagy sensitizes to olaparib. However, the concept that autophagy mediates resistance to olaparib is not necessarily novel and the proposed pathways have been previously demonstrated in other cancer models. The data is strictly correlative with respect to Rad51/Brca2, p62, and filamin A and their presumed dependency/relationship is based on associations previously reported in the literature. The lack of any in vivo data to support a treatment strategy also weakens the potential significance.
The majority of the manuscript is focused on comparing what happens when these cell lines are pretreated with rapamycin 1 day before olaparib vs treated 1 day after olaparib. The strongest data using this approach is demonstrating that rapamycin pretreatment causes a significant increase in cell proliferation, whereas delayed rapamycin pretreatment does not do this and KO of ATG16L1 abrogates this proliferative response, although to a lesser degree in PC3consistent with its increased resistance. The other highly supportive data is using an HR vs NHEJ reporter assay and finding that pre-treatment with rapamycin leads to increased HR, but not NHEJ activity. This HR response was blocked by loss of ATG16L1, thus linking autophagy specifically to the HR response.
1. There are several major concerns about the rest of the data. Unfortunately, the rest of the data is not consistent between the cell lines nor with repeat experiments shown in supplementary data. Same applies to the cell cycle analysis (Fig. 2D, S1C), where C4-2B seems to act as predicted, but the other lines are inconsistent.
While we agree that the three cell lines when compared to each other are distinct in the level of response, we would point out that the phenotype described is internally consistent within each of the cell lines, although the effect is less marked in the LNCaP and PC-3 cell lines. To make this point more clearly, we have modified the following sentence of the discussion in line 368: We hypothesize that this different level of autophagy-mediated resistance must be due to different factors such as AR expression, basal level of autophagy, or even the different degrees of autophagy activation by rapamycin. Fig 2B and S1B, rapamycin pretreatment, but not posttreatment leads to increased LC3II and decreased p62 in all lines. This is seen in LNCaP and C4-2B in Fig 2B, but only for LNCaP in Fig S1B and never for PC3. Furthermore, rapamycin or olaparib alone is often having the same impact as the combined treatment.

The authors claim that in
We thank the reviewer for this comment but we would like to note that we did not indicate that there is difference between LC3II and p62 between the pre and post treatment. Indeed, the following sentence can be found in the result section: "Both RO10 and O10R conditions showed similar levels of LC3-II, indicating similar levels of autophagy activation for day 2 and 6". For clarity we have modified the sentence as follows between line 151 and 153: No significant differences were observed for LC3-II expression in RO10 and O10R conditions indicating similar levels of autophagy activation (Fig. 2b and S2b). In addition, as there seems to be confusion around this point, we have added the quantification of LC3-II expression in Fig. 2c and S2b to clearly show that autophagy in pre-or post-activated populations is the same. It supports the notion that the effect we observed is due to the timeline of treatment and not different level of autophagy between these two conditions. While we agree that olaparib alone can have an impact on autophagy, in this study we are focusing on the impact of induced autophagy on the response to olaparib and therefore cell fates post-olaparib would be the same in all conditions.

Data in figures 3 and 5 don't include the post-treatment data to be able to directly
compare to the pre-treatment data within the same experiment. Instead the posttreatment data is placed with another set of data in supplementary figures 2,3, and 4. The quantification between these different sets of data leads to differences in results and differences in statistical significance. Interpretation was also complicated by the fact that Supplementary Fig 3 data seems to be superimposed over Supplementary Fig  2 data. This questions whether the stats presented are simply technical replicates within each experiment and not biological replicates.
In Figure 2 we included all experimental conditions so that the reader could appreciate the effect of the RO10 treatment. As the phenomena was consistent in subsequent figures, we chose to present the O10R treatments as supplementary material in order to unencumber the subsequent figures. As it was unclear that each experiment was a biological replicate we have now included this information in the supplementary figure legend. We regret the error in superimposing Supplementary Fig 2 and 3, we have resolved this issue in the resubmission.

Data in Figure 5C would be strengthened by demonstrating that p62 is also nuclear localized in these extracts. It is not clear why the quantification of the blots in Fig5C were normalized to olaparib rather than to untreated controls. The effect on filamin A is levels is modest at best. Quantification of several different blots from different experiments is required to assure statistical significance and consistency.
We thank the reviewer for this suggestion and agree that this information would strengthen our manuscript. We added a western blot tracking SQSTM1/p62 localization in the nucleus and the quantification from three independent experiments now are reported in Fig. 5C. We chose to compare RO10 and O10R to O10 alone as the main purpose of this experiment was to show how autophagy pre-activation/post-activation affects the nuclear level of FLNA and SQSTM1/p62 in the O10 condition. We added the following sentence in the results to address this comment between line 288 and 298:

O10R conditions or KO cell lines did not show this increase in FLNA expression under RO10
conditions. This increase of FLNA was accompanied by a significant decrease of SQSTM1/p62 in the nucleus in LNCaP WT (3.3-fold change), C4-2B WT (3.4-fold change) and PC-3 WT (4.5fold change) in RO10 conditions compared to O10 (Fig. 5c and e). This effect was not observed in O10R condition and in PC KO Atg16L1 cell lines.

Overall these results are only correlative with respect to Rad51/Brca2, p62, and filamin A and their presumed dependency/relationship. There are no experiments that disrupt expression/activity of any of these proteins and show that they impact the response to olaparib. Authors need to be careful about using strong statements indicating their data support their proposed model.
We thank the reviewer for this comment and agree that it would be beneficial to provide direct evidence for this relationship. We conducted supplementary experiments where we modulated p62 expression using a siRNA approach. This demonstrated that pre-siRNA knock-down of SQSTM1/p62 (siO10) in the PC KO Atg16L1 cell lines had similar effects as RO10 on cell proliferation ( Fig. 6c and S9a), on H2AX foci resolution ( Fig. 6d and S9b) and HR efficiency (Fig. 6e). Post-transfection (O10si) had no effects as O10R. In addition to these new figures we added the following text in the results between line 303 and 329:

Targeting SQSTM1/p62 rescued effect of autophagy in PC KO Atg16L1 cell lines
To confirm the importance of SQSTM1/p62 in this autophagy-mediated resistance, we used a siRNA against SQSTM1/p62 in our PC KO Atg16L1 cell lines and in WT ones ( Fig. 6 and   S9). We followed the sequence of treatment as rapamycin, by transfecting siRNA (si) or scramble (Sble) 24 hours before or after olaparib treatment (Sble/siO10, O10Sble/si; respectively) (Fig. S9a). We confirmed siRNA efficacity by western blot and observed an important decrease of SQSTM1/sip62 protein level mainly in PC KO Atg16L1 but also in PC WT cell lines at day 2 and 6 ( Fig. 6a). Interestingly, pre-inhibition of SQSTM1/p62 (siO10) in

. A similar decrease was observed in WT cells. In all Sble and
O10si conditions, levels of γ-H2AX foci were similar from O10 conditions for all cell lines (Fig.   S9c). To determine if this decrease in γ-H2AX foci was due to an increase of HR efficiency, we used our GFP reporter assay (Fig. 6d) 6. The authors should also acknowledge that blocking ATG16L1 or autophagy in general will result in p62 accumulation simply because you are blocking its degradation.
We agree and clarified this point as follow in line 270: "As expected, we observed that the KO cell lines had higher expression levels of SQSTM1/p62 compared to WT cells (Fig. 2B)."

It should be noted that C4-2B and LNCaP display a dramatic response to olaparib (logfold) upon autophagy inhibition, but PC-3 displays less than a 2-fold increase in sensitivity, indicating at least for PC-3 cells, there are other factors involved. The authors need to acknowledge and address this more in the Discussion.
Clarity around this point has already been addressed in our response to Reviewer 1.

Reviewer #3 (Remarks to the Author):
Cahuzac et al investigated the impact of autophagy on prostate cancer cell response to the PARPi olaparib by following the autophagy activation timeline. And Cahuzac et al define that pre-activation of autophagy before olaparib treatment reduces the level of SQSTM1/p62 in the nucleus, results in an increase of DNA repair activity by homologous recombination to repair double-strand breaks induced by olaparib, and enhances cell proliferation. The authors should design and perform more experiments to support their conclusions.
1. The authors showed that PC-3 with a higher basal level of autophagy is resistant to olaparib compared to LNCaP and C4-2B with a lower basal level of autophagy (Fig. 1A-C). How about the basal expression level of PARP1/2, which may also affect the antiproliferation effect of olaparib?
We thank the reviewer for raising an interesting point. We addressed this by performing an additional western blot (Fig. S1) to determine the expression of PARP1 and PARylation. We observed that depletion of autophagy did not impact PARP1 expression and PARylation. The following sentence has been added to the results to address this point between line 140 and

143:
To ensure that PARP1 or PARylation did not affect olaparib sensitivity, we measured basal level protein by western blot and found no differences significant difference between WT PC-3 cells and those undergoing autophagy depletion (Fig. S1).

The authors showed that AR-negative PC cells may have a higher basal level of autophagy and are olaparib-resistant compared to AR-positive PC cells (Fig. 1A-C). But the authors only chose one AR-negative PC cell line PC-3, and also didn't include 22Rv1 cell line that is AR-V7-positive. How about the basal level of autophagy and the IC50 values for olaparib in PC-3 and 22Rv1 cells?
We have unpublished results that measure basal level of autophagy (western blot) and olaparib sensitivity (clonogenic) in the 22Rv1 cell line. We observed similar levels of autophagy as LNCaP and C4-2B and 22Rv1 cells had an olaparib IC50 of 0.94 µM. As AR-V7 expression could be a confounding variable, we decided not to continue with this cell line. Fig. 2B). The cell proliferation was significantly increased under RO10 conditions, but not under O10R conditions (Fig. 2C). Even under O10R conditions, the proliferation of LNCaP, C4-2B and PC-3 cells was decreased compared to olaparib treatment alone (Fig. 2C). The authors should perform another experiment (may choose cell viability assay using CellTiter-Glo) to further confirm the different anti-proliferation effect under RO10 and O10R conditions, which both showed a similarly high level of autophagy activation.

The authors mentioned "Both RO10 and O10R conditions showed similar levels of LC3-II, indicating similar levels of autophagy activation" (
We again thank the reviewer for this comment but we feel that our choice of proliferation assay is more judicious based on the phenomena we are trying to characterize. More specifically, we know that autophagy impact metabolism (production of ATP, mitochondrial activity (Guo, Teng et al. 2016, Ferro, Servais et al. 2020), and therefore the Incucyte system, which directly visualizes and quantifies proliferation, is more appropriate. Fig.2B, the western blots showed that in C4-2B WT cells, the autophagy was not activated, even was inhibited in all the groups with Rapamycin treatment, compared to the Control group as shown by LC3-II and p62. The authors should further confirm that. For these western blots, the authors may also show the normalized values or plots explaining the change in expression normalized to Actin.

In
We agree and have added the quantification of LC3-II in Fig.2c and S2b to clarify the visualization of data. Fig. 2C, it showed that the olaparib treatment (O10) inhibited the proliferation of PC-3 Atg16L1 KO cells more than that of PC-3 WT cells. But in Fig. 1G-H, the authors have demonstrated that complete depletion of autophagy reduced the olaparib IC50 value from2.13 μM to 1.5 μM for PC-3. The authors should further confirm that.

In
While the observation the reviewer makes is correct, we would point out that the IC50 results are based on clonogenic assays which are difficult to directly compare to proliferation assays.
However, the critical point is that the overall trend is consistent. In both cell proliferation and IC50 experiments we observed a decrease of these two parameters when autophagy was abrogated.
6. The authors showed that pre-activation of autophagy limited the effect of olaparib on cell proliferation and cell cycle and this protective effect was abrogated when autophagy was abrogated by Atg16L1 KO (Fig.2). The authors should do one more knockout rescue experiment to see if the re-activation of autophagy would make the PC cells sensitive to olaparib again.
We agree that this result would strengthen the manuscript. We therefore confirmed our results by a rescue experiment as suggested using Atg16L1-HA and we reperformed cell proliferation assays. These results are now presented in Fig. S3 where we demonstrate that the rescue of Atg16L1 expression restore dynamics of autophagy and where RO10 conditions had the same effect on cell proliferation in this PC KO-rescue Atg16L1 as PC WT. We also did the experiment in WT-rescue Atg16L1 to ensure that rescue did not affect our previous observations. We also added the following text in the results between line 159 and 166: We also rescued the depletion of autophagy by introducing a plasmid coding for Atg16L1 with a HA-tag in our PC WT and KO Atg16L1 cell lines (Fig. S3). Expression of Atg16L1-HA restored autophagy dynamics by the lipidation of LC3-I in LC3-II that was not observed in PC KO cell lines in autophagy induction and inhibition conditions ( Fig. S3a and Fig. 1f) S3b and Fig. 2d).
7. The authors showed that using their plasmid-based DNA repair reporter assays, complete depletion of autophagy by Aty16L1 KO significantly reduced the efficiency of HR (Fig.4F). Could the efficiency of HR be rescued with re-expression of Aty16L1?
We confirmed our results by a rescue experiment as suggested using Atg16L1-HA and we reperformed the measurement of DNA repair efficiency. These results can be found in Fig. S7 where  (Fig. S7). Following the reviewer suggestion, new western blots were performed where PC WT and PC KO were loaded on the same gel (new Fig. 2b). As the visualization of difference is clear we felt we did not have to add additional plots to clearly demonstrate the trend. Fig. 5C, it seems the loadings of samples are not equal as shown by loading control SP1 and tubulin. The authors should normalize the bands of FLNA to loading control SP1 or tubulin. And the authors should also include SQSTM1/p62 in Fig. 5C to visually show the change of FLNA along with p62 in the nucleus.

In
We have taken into account this comment. SP1 was used as a control of the nuclear fraction and tubulin as a control for the cytoplasmic fraction. We added SQSTM1/p62 expression in the nucleus in Fig. 5c and the quantification of FLNA and SQSTM1/p62 in Fig. 5d-e. As the nuclear fraction is the most critical, we have moved the cytoplasmic fraction to Fig. S8c and kept SP1, tubuline and Atg16L1 as controls. The following text has been added to the results: We determined whether FLNA was important in this autophagy-mediated resistance to olaparib by measuring the nuclear fraction of FLNA and SQSTM1/p62 (Fig. 5c-d  and PC-3 WT (4.5-fold change) in RO10 conditions compared to O10 (Fig. 5c and e). This effect was not observed in O10R condition and in PC KO Atg16L1 cell lines.
We are sorry for the confusion, stripping of Atg16L1 blot was not sufficient to remove all Atg16L1 in these two cell lines. The problem was solved in a new western blot.
b. In Fig. S2, there is no panel C and D. Both Fig. S2C and S2D are missed. The whole Fig.S3 is missed.
We are sorry about the superimposition between Supplementary Fig 2 and 3, we have resolved this issue in the resubmission.