Deregulated PP1α phosphatase activity towards MAPK activation is antagonized by a tumor suppressive failsafe mechanism

The mitogen-activated protein kinase (MAPK) pathway is frequently aberrantly activated in advanced cancers, including metastatic prostate cancer (CaP). However, activating mutations or gene rearrangements among MAPK signaling components, such as Ras and Raf, are not always observed in cancers with hyperactivated MAPK. The mechanisms underlying MAPK activation in these cancers remain largely elusive. Here we discover that genomic amplification of the PPP1CA gene is highly enriched in metastatic human CaP. We further identify an S6K/PP1α/B-Raf signaling pathway leading to activation of MAPK signaling that is antagonized by the PML tumor suppressor. Mechanistically, we find that PP1α acts as a B-Raf activating phosphatase and that PML suppresses MAPK activation by sequestering PP1α into PML nuclear bodies, hence repressing S6K-dependent PP1α phosphorylation, 14-3-3 binding and cytoplasmic accumulation. Our findings therefore reveal a PP1α/PML molecular network that is genetically altered in human cancer towards aberrant MAPK activation, with important therapeutic implications.

In this paper the authors report the positive regulation of ERK signalling by PP1a. Investigating the mechanism they find that PP1a is sequestered in nuclear PML bodies, from where it can be released by S6K mediated phosphorylation and binding to 14-3-3g. It then can translocate to the cytosol to activate BRAF by dephosphorylation of inhibitory residues. This is an interesting finding. Although the regulation of RAF kinases by PP1 and PP2 has been reported before (Abraham et al., 2000;Dhillon et al., 2002;Jaumot and Hancock, 2001;Strack, 2002), the aspect of the spatial regulation is new and adds to our understanding of the regulation of this critical pathway. However, the study is superficial and somewhat preliminary in parts, and quantitation of the results is largely missing. These shortcomings should be amended before publication can be recommended. Specific comments The critical role of PP1 and PP2A mediated dephosphorylation in the activation of CRAF and BRAF kinases has been reported before (Abraham et al., 2000;Dhillon et al., 2002;Jaumot and Hancock, 2001;Strack, 2002), and should be cited and discussed. The hypothesis that PML-NBs sequestrate PP1a and thereby restrict ERK activity needs to be further corroborated, e.g. by mapping the binding sites of the PML-PP1a interaction and showing that binding site mutants activate ERK; or showing that a PP1a constitutively targeted to the nucleus via a nuclear localisation signal does not activate ERK. The role of S6K should be further elaborated. Does knockdown of S6K prevent the activation of ERK by PP1a? Also, as S6K is a downstream target of the ERK pathway it potentially could serve as part of a positive feedback loop to regulate the activation kinetics of ERK? Is this the case? Does the interaction of PP1a with PML only sequester ERK or does it also affect the PP1a phosphatase activity? This could be tested for instance by measuring PML associated phosphatase activity or by adding purified PML to PP1a phosphatase assays in vitro. What is the ERK activity in the nuclear fraction plus/minus PML knockdown? Is 14-3-3g only needed for the cytoplasmic localisation of PP1a or does it also regulate its enzymatic activity? Figs. 2A and S1A. The activation of ERK by PP1A overexpression is pronounced in HEK293, but rather subtle in LNCaP cells. To obtain a reliable indication of the effect, a quantitation obtained from at least 3 independent experiments should be shown. Figs. S1C,D. The effect of PP2A is complex, as it can activate RAF by dephosphorylation of inhibitory residues as well as de-activate MEK and ERK by dephosphorylation of activating residues. Thus, measuring ppERK is not sufficient to assess the effects of PP2A. These experiments should be complemented by RAF in vitro kinase assays. Fig. 2f. The change in ERK activity caused by the different knockdowns is rather small (20% or less in most conditions). These experiments should be quantitated and shown as bar graphs with error bars. Fig. 4B. The effect of mutating the ERK feedback phosphorylation sites in BRAF is rather subtle. The results should be quantified, and the quantification shown as bar graph with error bars. Given the small effects of mutating the ERK sites, the authors also should consider mutating the other known inhibitory phosphorylation sites in BRAF, i.e. S365, S429, T440 (Guan et al., 2000); S465/467 (Holderfield et al., 2013);S614 (Dernayka et al., 2016) Minor comments Fig. 2B. The labelling of the upper panel is incomplete. Fig. S1E. How much lysate was loaded in the "input" lane compared for what was used for the IP? Fig. S1G. Tautomycin is not a PP1a specific inhibitor. It inhibits PP1 and PP2 almost with equal potency (Hori et al., 1991).
Chen and colleagues present biochemical evidence that MAPK pathway may become aberrantly activated in CaP due to PPP1CA (the catalytic subunit of PP1a) genomic amplifications and/or by a non-genomic S6K/PP1a/B-Raf signaling pathway. In the latter scenario, mutations of PML, which normally functions to sequester PP1a into NBs, lead to the accumulation of PP1a in the cytoplasm to dephosphorylate several inhibitory phosphor-sites on B-Raf kinase resulting in MAPK activation. This is an interesting and strong biochemistry study that sheds novel light on how the MAPK pathway, in the absence of component mutations, might become hyper-activated in cancer cells especially CaP cells.
1. One of the gaps in our knowledge from this study is on biology. For example, since PPP1CA is amplified in 17-25% mCRPC, the overexpressed PP1a may promote CaP cell invasion. And if this effect is mediated, at least in part, via activation of B-Raf kinase and downstream MAPK, the pathway inhibitors should blunt PP1a-promoted CaP invasion. 2. In the scenario of increased PP1a activity due to co-deletion of PTEN/PML (without PPP1CA amplification), some quantitative information may help readers appreciate the significance of the proposed signaling pathway. That is, in normal cells (with normal levels of PTEN/PML), how much of PP1a is in the cytoplasm versus sequestered in the PML NBs? How is the protein redistributed in the two cellular compartments in the absence of PML? 3. For data in Figure 2f: Ideally, the authors should repeat the experiment several times and present quantitative data in a bar graph for the changes in p-ERK. 4. How did acute loss of PTEN lead to upregulation of PML? 5. When referring to 'the phosphatase-inactive PP1a mutant (H248K) (page 6), a reference should be provided.

Rebuttal to the Reviewers
We thank both of the Reviewers for their constructive and positive comments that have helped significantly improve the quality, structure and content of this manuscript.
We are pleased that we were able to address each specific comment in full, as outlined in the point-by-point rebuttal section below. Importantly, based on the Reviewers comments and suggestions, we have tremendously improved our study by performing multiple new experiments with both quantitative and qualitative data, as presented in this new version of the manuscript.
We hope that our revised manuscript will now meet the satisfaction of the Reviewers and be deemed suitable for publication in Nature Communications.
A detailed rebuttal to each Reviewer's specific comments is included below.

Referee #1 (Remarks to the Author):
In this paper the authors report the positive regulation of ERK signalling by PP1a. Investigating the mechanism they find that PP1a is sequestered in nuclear PML bodies, from where it can be released by S6K mediated phosphorylation and binding to 14-3-3g. It then can translocate to the cytosol to activate BRAF by dephosphorylation of inhibitory residues. This is an interesting finding. Although the regulation of RAF kinases by PP1 and PP2 has been reported before (Abraham et al., 2000;Dhillon et al., 2002;Jaumot and Hancock, 2001;Strack, 2002), the aspect of the spatial regulation is new and adds to our understanding of the regulation of this critical pathway. However, the study is superficial and somewhat preliminary in parts, and quantitation of the results is largely missing. These shortcomings should be amended before publication can be recommended.
We thank the Reviewer for recognizing the novelty of our study and for his/her constructive comments regarding our manuscript. In addressing each of the Reviewers concerns, we have been able to bring additional and more insightful data to the manuscript. We are confident that the Reviewer will appreciate the fact that our conclusion has now been significantly strengthened by these additional quantitative and qualitative data.

Specific comments
The critical role of PP1 and PP2A mediated dephosphorylation in the activation of CRAF and BRAF kinases has been reported before (Abraham et al., 2000;Dhillon et al., 2002;Jaumot and Hancock, 2001;Strack, 2002), and should be cited and discussed.
We apologize for this oversight. We have now included and discussed these references suggested by the Reviewer. They can now be found on Page 11 of our revised manuscript (Line #9).
The hypothesis that PML-NBs sequestrate PP1a and thereby restrict ERK activity needs to be further corroborated, e.g. by mapping the binding sites of the PML-PP1a interaction and showing that binding site mutants activate ERK; or showing that a PP1a constitutively targeted to the nucleus via a nuclear localisation signal does not activate ERK.
We thank the Reviewer for this comment. As suggested by the Reviewer, we have generated a PP1α mutant constitutively targeted to the nucleus by fusing the NLS sequence derived from c-Myc (Ray et al. Bioconjug Chem 2015, Fig . 1a) to C-terminus of PP1α. Indeed, compared to wild type PP1α, the NLS-PP1α displayed a drastically decreased capacity to activate ERK. These results further corroborate the role of subcellular compartmentalization in PP1α-induced ERK activation and are now shown in our new Fig. 2e.

The role of S6K should be further elaborated. Does knockdown of S6K prevent the activation of ERK by PP1a?
We have further investigated the role of S6K in PP1α-induced ERK activation and have found that knockdown of S6K1 via small interfering RNA largely suppresses PP1α-induced ERK activation. This again confirms that the activation of ERK by PP1α is dependent on S6K. These data are now shown in our new Fig. 3e.
Also, as S6K is a downstream target of the ERK pathway it potentially could serve as part of a positive feedback loop to regulate the activation kinetics of ERK? Is this the case?
The Reviewer raises an excellent point that S6K-PP1α-B-Raf-ERK could form a feed-forward loop supporting sustained ERK activation. Based on our current and previous studies (Ma et al. Cell 2005), this could be true in the context of PML loss or PPP1CA amplification. However, we and others have previously also shown that AKT/mTOR/S6K activation triggers a negative feedback on MAPK signaling pathway, presumably as a result of the upstream IRS inactivation induced by S6K (Shah et al. Curr Biol 2004;Carracedo et al. J Clin Invest 2008) (see also our Supplementary Fig. 2). Thus, depending on the genetic context, S6K is a double-edge sword in the regulation of MAPK signaling, since it can act as both a suppressor of ERK activation, in the context of intact PML function, and as a promoter of sustained ERK activation, in the context of PML loss or PPP1CA amplification. We have now also included this important discussion on Page 12 of our revised manuscript (Line #2).

Does the interaction of PP1a with PML only sequester ERK or does it also affect the PP1a phosphatase
activity? This could be tested for instance by measuring PML associated phosphatase activity or by adding purified PML to PP1a phosphatase assays in vitro.
We thank the Reviewer for this insightful comment. To determine whether PML affects the PP1α phosphatase activity, we have now performed in vitro PP1α phosphatase assays towards its nuclear targets. The transcription factor cAMPresponsive element binding protein (CREB) is a PP1α target in the nucleus, where PP1 dephosphorylates CREB at Ser133 and inhibits CREB-mediated transcriptional activation (Hagiwara et al. Cell 1992). We have found that purified GST-PML does not affect the CREB dephosphorylation exerted by PP1α in vitro, suggesting that PML primarily functions as a scaffold/chaperone for PP1α rather than as a direct regulator of the PP1α phosphatase activity. These data are now shown in our new Supplementary Fig. 1m.

What is the ERK activity in the nuclear fraction plus/minus PML knockdown?
To address the Reviewer's comment, we have examined the transcriptional activity of the Elk-1 transcription factor, a well-established ERK nuclear target (Marais et al. Cell 1993), in the absence and in the presence of PML knockdown. We have found that knockdown of PML results in the activation of the Elk reporter to a greater extent than the siRNA control ( Fig. 1 for the Reviewer). This is consistent with our complementary studies demonstrating that PML loss induces ERK activation (Chen and Pandolfi et al. Nat Genet 2017, manuscript in revision).

Is 14-3-3g only needed for the cytoplasmic localisation of PP1a or does it also regulate its enzymatic activity?
We thank the Reviewer for this critical comment. To determine whether 14-3-3-γ regulates the PP1α phosphatase activity, we have performed in vitro phosphatase assays and have found that 14-3-3-γ has no effect on the PP1α phosphatase activity towards dephosphorylating B-Raf kinase. These results are now shown in our new Supplementary Fig. 1n.
Figs. 2A and S1A. The activation of ERK by PP1A overexpression is pronounced in HEK293, but rather subtle in LNCaP cells. To obtain a reliable indication of the effect, a quantitation obtained from at least 3 independent experiments should be shown.
We thank the Reviewer for this comment. Given that LNCaP cell line is one of the hard-to-transfect cell lines (Fronsdal et al. Prostate 2000), the main reason for the observed difference in the activation of ERK by PP1α between 293T and LNCaP cells, we believe, is the lower transfection efficiency of PP1α in LNCaP cells. As suggested by the Reviewer, we have now performed three independent transfection experiments using a better transfection reagent, TransIT-X2 from Mirus Bio LLC, rather than lipofectamine 2000. Along with improved transfection in LNCaP, PP1α is now able to activate ERK more than six-fold compared to the empty vector control. The representative results, plus quantitation data, are now shown in a revised Supplementary Fig. 1a.

Figs. S1C,D. The effect of PP2A is complex, as it can activate RAF by dephosphorylation of inhibitory residues as well as de-activate MEK and ERK by dephosphorylation of activating residues. Thus, measuring ppERK is not sufficient to assess the effects of PP2A. These experiments should be complemented by RAF in vitro kinase assays.
The Reviewer's comments here are well taken. As suggested by the Reviewer, we have now performed in vitro kinase assays to show the effect of PP2A on Raf kinase activity. Indeed, PP2A positively regulates B-Raf kinase activity towards phosphorylating MEK. Interestingly, we have also found that PP1α is a more potent activator of B-Raf than PP2A. These data are now shown in our new Supplementary Fig. 1b. Additionally, according to the Reviewer's comments, we have now included a statement to highlight the complex role of PP2A in the regulation of the MAPK signaling. They can now be found on Page 5 of our revised manuscript (Last line).

Fig. 2f. The change in ERK activity caused by the different knockdowns is rather small (20% or less in most conditions). These experiments should be quantitated and shown as bar graphs with error bars.
We agree with the Reviewer and, thus, have now repeated the experiment three times. The representative results, plus quantitation data in a bar graph, are now included in a revised Fig. 2f that now appears as Fig. 2g.

Fig. 4B. The effect of mutating the ERK feedback phosphorylation sites in BRAF is rather subtle. The results should be quantified, and the quantification shown as bar graph with error bars.
We agree with the Reviewer and, thus, have now repeated the experiment three times. The representative results, plus quantitation data in a bar graph, are now included in a revised Fig. 4b.
Following the insightful comments from the Reviewer, we have now investigated the effect of PP1α on other known inhibitory phosphorylation sites suggested by the Reviewer. We have confirmed that the B-Raf mutants devoid of these inhibitory sites indeed exhibit an enhanced ability to activate MAPK signaling. However, they are all still sensitive to PP1α activation. These results further corroborate that PP1α appears to exert its effect on B-Raf primarily through the ERK-regulated inhibitory sites and are now shown in our new Supplementary Fig.  1i-k. We thank the Reviewer for pointing this out. The missing labels in the upper panel of Fig. 2b are now included.   (Hori et al., 1991).

Minor comments
The Reviewer is correct that tautomycin is not a PP1α specific inhibitor. We apologize for this oversight. Given that tautomycin does display a slight preference for PP1 inhibition relative to PP2A inhibition in multiple studies (MacKintosh et al. FEBS Lett 1990, Fig . 2; Favre et al. J Biol Chem 1997, Fig . 7a; Swingle et al. Methods Mol Biol 2007 , Table 1), we have now changed the sentence into: "In line with this, PC3 cells treated with tautomycin, a more selective inhibitor for PP1α, displayed a dose-dependent inhibition of EGF-induced ERK phosphorylation". We have also cited the references mentioned in this comment by the Reviewer and ourselves.
Chen and colleagues present biochemical evidence that MAPK pathway may become aberrantly activated in CaP due to PPP1CA (the catalytic subunit of PP1a) genomic amplifications and/or by a non-genomic S6K/PP1a/B-Raf signaling pathway. In the latter scenario, mutations of PML, which normally functions to sequester PP1a into NBs, lead to the accumulation of PP1a in the cytoplasm to dephosphorylate several inhibitory phosphor-sites on B-Raf kinase resulting in MAPK activation. This is an interesting and strong biochemistry study that sheds novel light on how the MAPK pathway, in the absence of component mutations, might become hyper-activated in cancer cells especially CaP cells.
We are extremely pleased that the Reviewer regards our study as interesting as well as strong, and we thank the Reviewer for his/her encouraging review. We also would like to thank the Reviewer for the constructive criticisms that he/she has outlined below. We thank the Reviewer for this critical comment. To address these points, we have established CaP cell lines, stably overexpressing PP1α, and have used them for cell migration and invasion assays in the absence and in the presence of the MEK inhibitor, U0126. We have found that CaP cells stably overexpressing PP1α exhibit higher ERK activation along with significantly increased cell migration and invasion. Notably, treatment with the MEK inhibitor, U0126, in CaP cells represses not only basal but also PP1α-induced cell migration and invasion. These functional data, together with the human genetic and mechanistic analyses, implicate PPP1CA as a prometastatic proto-oncogene in human CaP and MAPK signaling as one of the key downstream effectors of PP1αinduced cell invasiveness. These important new results are now shown in our new Fig. 4e and Supplementary  Fig. 1l. 2. In the scenario of increased PP1a activity due to co-deletion of PTEN/PML (without PPP1CA amplification), some quantitative information may help readers appreciate the significance of the proposed signaling pathway. That is, in normal cells (with normal levels of PTEN/PML), how much of PP1a is in the cytoplasm versus sequestered in the PML NBs? How is the protein redistributed in the two cellular compartments in the absence of PML?
We thank the Reviewer for this great suggestion. To quantify the changes in the cellular distribution of PP1α upon co-loss of PTEN/PML, we have now repeated the WI-38 cellular fractionation assays in Fig. 2d three times. WI-38 cells are normal human diploid fibroblasts with intact PTEN and PML protein expression. We have found that PP1α primarily localizes in the nucleus (80% nuclear and 20% cytoplasmic) in WI-38 control cells, whereas the cellular distribution of PP1α was reversed upon knockdown of PTEN and PML (30% nuclear and 70% cytoplasmic). The representative results, plus quantitation data in a bar graph, are now included in a revised Fig. 2d. Figure 2f: Ideally, the authors should repeat the experiment several times and present quantitative data in a bar graph for the changes in p-ERK.

For data in
We agree with the Reviewer and, thus, have now repeated the experiment three times. The representative results, plus quantitation data in a bar graph, are now included in a revised Fig. 2f that now appears as Fig. 2g.