Author response:
The following is the authors’ response to the original reviews.
Public Reviews:
Reviewer #1 (Public review):
Summary:
This manuscript reports the discovery and characterization of the first bifunctional degrader of tankyrase. Notably, the tankyrase degrader exhibits stronger β-catenin inhibition and tumor growth suppression compared to conventional tankyrase inhibitors. Mechanistically, while tankyrase inhibitors stabilize tankyrase and promote Axin puncta formation - thereby impairing β-catenin degradation - the degrader avoids this effect, resulting in deeper suppression of β-catenin signaling. These findings suggest that targeted degradation of tankyrase offers a novel therapeutic strategy for β-catenin-driven cancers. Overall, this is a compelling study with significant translational potential.
Strengths:
(1) The manuscript presents a rigorous and well-executed study on a timely and impactful topic.
(2) The biochemical and cellular characterization of the tankyrase degrader is thorough, and the comparative analysis with tankyrase inhibitors is insightful.
(3) The finding that tankyrase stabilization by inhibitors may interfere with Axin function is novel and significant. It aligns with earlier observations (e.g., Huang 2009) that transient tankyrase overexpression can stabilize β-catenin independently of PAR domain activity.
(4) The use of TNKS1/2 knockout cells expressing catalytically inactive tankyrase to demonstrate β-catenin inhibitory activity of the tankyrase degrader is elegant.
(5) The finding that the tankyrase degrader has superior anti-proliferative effects in colorectal cancer models has important therapeutic implications.
Weaknesses:
(1) A key caveat is that the identified tankyrase degrader also targets GSPT1 for degradation. This raises the possibility that GSPT1 degradation may contribute to the observed β-catenin and tumor growth inhibition.
(2) The authors address this concern reasonably by showing that DLD1 cells resistant to GSPT1 degradation remain sensitive to the tankyrase degraded.
(3) To further strengthen this point, the authors might consider generating TNKS1/2 double knockout cells (e.g., in DLD1 or SW480 backgrounds) and demonstrating that the degrader loses its growth-inhibitory effect in these models. However, given the technical challenges of creating double knockouts in cancer cell lines, such experiments could be considered optional.
We thank the Reviewer for the favorable feedback. The major concern is the collateral degradation of GSPT1. As the Reviewer noted, IWR1-POMA was able to suppress colony formation in DLD-1 cells resistant to a GSPT1/2 degrader (DLD-1R, Figure 6B and S9F), suggesting that TNKS but not GSPT degradation is responsible for growth inhibition.
We also appreciate that the Reviewer brought it to our attention an important early observation of the TNKS scaffolding effects. Cong reported in 2009 that overexpression of TNKS induced AXIN puncta formation in a SAM but not PARP domain-dependent manner (PMID: 19759537, Ref. 12). We have added this reference to the introduction of TNKS scaffolding in the revised manuscript.
Reviewer #2 (Public review):
Summary:
The ADP-ribosyltransferase tankyrase controls many biological processes, many of which are relevant to human disease. This includes Wnt/beta-catenin signalling, which is dysregulated in many cancers, most notably colorectal cancer. Tankyrase is a positive regulator of Wnt/beta-catenin signalling in that it counters the activity of the beta-catenin destruction complex (DC). Catalytic inhibition of tankyrase not only blocks PAR-dependent ubiquitylation and degradation of AXIN1/2, the central scaffolding protein in the DC, but also tankyrase itself. As a result, blocking tankyrase gives rise to tankyrase accumulation, which may accentuate its non-catalytic functions, which have been proposed to drive Wnt/beta-catenin signalling. Most tankyrase catalytic inhibitors have shown limited efficacy and substantial toxicity in vivo. By developing tankyrase-directed PROTACs, the authors aim to block both catalytic and non-catalytic functions of tankyrase, aspiring to achieve a more complete inhibition of Wnt/beta-catenin signalling. The successfully developed PROTAC, based on the existing catalytic inhibitor IWR1, IWR1-POMA, induces the degradation of both TNKS and TNKS2, blocks beta-catenin-dependent transcription without stabilising the DC in puncta/degradasomes, and inhibits cancer cell growth in vitro. Mechanistically, this points to a scaffolding role of tankyrase in the DC, at least under conditions of tankyrase catalytic inhibition, in line with previous proposals.
Strengths:
The study clearly illustrates the incentive for developing a tankyrase degrader, namely, to abolish both catalytic and non-catalytic functions of tankyrase. By and large, the study achieves these ambitions, and the findings support the main conclusions, although the statement that a more complete inhibition of the pathway is achieved requires corroboration. The proteomics studies are powerful. IWR1-POMA constitutes a very useful tool to re-evaluate targeting of tankyrase in oncogenic Wnt/beta-catenin signalling. The paired compounds will benefit investigations of tankyrase scaffolding functions across many different biological systems controlled by tankyrase. The findings are exciting.
Weaknesses:
Although the results are promising and mostly compelling, the claim that the PROTACs provide "a deeper suppression of the WNT/β-catenin pathway activity" requires further corroboration, particularly at endogenous tankyrase levels.
We thank the Reviewer for the encouraging and insightful comments. The major critique concerns whether TNKS degraders can suppress WNT/β-catenin signaling more effectively than TNKS inhibitors at endogenous TNKS levels. IWR1-POMA reduced the level of cytosolic β-catenin more effectively than IWR1 in Wnt3A-stimulated HEK293 cells without protein overexpression (Figure 1D). IWR1POMA also suppressed STF activity more effectively than IWR1 in DLD-1 cells (Figure S8C) and reduced the expression levels of several WNT/β-catenin targets more effectively than IWR1 (Figure 1G and S8D). These results support that TNKS degraders can suppress WNT/β-catenin signaling more effectively than TNKS inhibitors at endogenous TNKS levels.
There are also some other points that, if considered, would further improve the manuscript, as detailed below.
(1) Abstract and line 62: Many catalytic tankyrase inhibitors tend to display toxicity, which is likely on-target (e.g., 10.1177/0192623315621192; 10.1158/0008-5472). This constitutes the main limiting factor for these compounds. An incomplete inhibition of Wnt/beta-catenin signalling may contribute to the challenges, but this does not appear to be the dominant problem. A more prominent introduction to this important challenge is probably expected by the field.
A previous study showed that G007-LK, a selective TNKS inhibitor, exhibited weak efficacy and dose-limiting toxicity at 5‒30 mg/kg BID or 10‒60 mg/kg QD in various mouse xenograft models (PMID: 23539443, Ref. 28). Similarly, G-631, another TNKS inhibitor, also showed dose-limiting toxicity without significant efficacy at 25‒100 mg/kg QD in mice (PMID: 26692561, Ref. 60). However, other studies showed that G007-LK was well-tolerated at 200 mg/kg QD over 3 weeks in mice (PMID: 29316982, Ref. 61), and treating mice with G007-LK at 10 mg/kg QD over 6 months also improved glucose tolerance without notable toxicity (PMID: 26631215, Ref. 62). Importantly, basroparib, a selective TNKS inhibitor, was well tolerated in a recent clinical trial (PMID: 40964966, Ref. 64), and constitutive silencing of both TNKS1 and TNKS2 for 150 days in APC-null mice prevented tumorigenesis without damaging the intestines (PMID: 31337618, Ref. 8). We have included some discussion of the toxicity issue associated with TNKS targeting at the end of the Discussion section.
(2) The authors do a good job in setting the scene for the need for tankyrase degraders. Their observations relating to the formation of puncta (degradasomes) being tankyrase-dependent are compatible with a previous study by Martino-Echarri et al. 2016 (10.1371/journal.pone.0150484): simultaneous silencing of TNKS and TNKS2 by RNAi abolishes degradasome formation. The paper is cited as reference 17, but only in passing, and deserves more prominence. (It includes an entire paragraph titled "Expression of tankyrases 1 and 2 is required for TNKSi-induced formation of axin puncta").
Indeed, Henderson’s 2016 paper (PMID: 26930278, previously Ref. 17, now Ref. 18) shed important light on the role of TNKS scaffolding in the DC. However, whereas this study demonstrated that knocking down both TNKS1 and TNKS2 by siRNA prevented G007-LK to induce AXIN puncta, it concluded that “puncta formation requires both the expression and the inactivation of TNKS,” which is inconsistent with our observations that accumulation of either catalytically active or inactive TNKS can promote AXIN puncta formation. The function roles of TNKS scaffolding in the DC also remained unaddressed. We have included additional discussion of Henderson’s findings in the first paragraph the Discussion section.
(3) Moreover, the scaffolding concept has been discussed comprehensively in other studies: 10.1111/bph.14038 and more recently 10.1042/BCJ20230230. There are also a few studies that focus on targeting the ankyrin repeat clusters of tankyrase to disengage substrates (10.1038/s41598-020-69229-y; 10.1038/s41598-019-55240-5) that illustrate the concept of blocking the scaffolding function. In that sense, the hypotheses are mature, and it is interesting to see some of them supported in this study. The authors could improve how they set their work into the context of these other efforts and proposals.
Indeed, Guettler demonstrated in 2016 that TNKS scaffolding could promote WNT/β-catenin signaling, which forms the basis of the current work. Meanwhile, whereas there have been efforts to target the SAM or ARC domain to address TNKS scaffolding by Guettler and Lehtiö, our approach of targeting TNKS for degradation is complementary. We have included in the last paragraph of the Discussion section information on efforts to target the ARC or SAM domains as an alternative approach to suppress WNT/β-catenin signaling without promoting TNKS oligomerization (PMID: 31836723 and 32704068, Ref. 66 and 67).
(4) In several places in the manuscript, the DC is referred to as "biomolecular condensate", at times even as a "classic example", implying that it operates through phase separation. This has not been demonstrated. In fact, super-resolution microscopy indicates that the puncta are not droplet-like (10.7554/eLife.08022), which would argue against the condensate hypothesis.
Biomolecular condensates are membraneless cellular compartments formed by phase separation of biomolecules, regardless of their physical/material properties (PMID: 28935776 and 28225081, Ref. 22 and 23). Super-resolution microscopy studies by Stenmark (PMID: 26124443, Ref. 17) showed that AXIN, APC, TNKS, and β-catenin interacted with each other to assemble into membraneless complexes, wherein AXIN and APC formed filaments throughout the DC. Peifer has also summarized evidence that supports the condensate nature of the DC (PMID: 30782412, Ref. 9; see also PMID: 26393419). However, we acknowledge that testing the physical properties of reconstituted DC (for example, PMID: 34352208) with TNKS will provide a better understanding of the nature, for example liquid vs. gel, of these condensates.
(5) It is beautiful to be able to use IWR1 and IWR1-POMA at identical concentrations for direct comparisons. However, this requires the two compounds to bind to tankyrase similarly well and reach the target to a comparable extent. How sure are authors that target engagement is comparable? Has this been evaluated?
Using a BRET assay, we have confirmed that IWR1-POMA binds to TNKS1 with affinity comparable to that of IWR1. Details of this study is now included in the Results sections, and the data are presented in the Supplementary Information (Fig. S3E–G).
(6) Figure 1F: It is not immediately apparent how IWR1-POMA shows more complete containment of Wnt/beta-catenin signalling. Most Wnt/beta-catenin targets lie close to the perfect diagonal, so I do not see how the statement "that IWR1-POMA controlled WNT/β-catenin signaling more effectively than IWR1" (in the legend of Figure 1F) is supported. Minimally, an expanded explanation would benefit the reader. Providing the colour-coding legend directly in the figure would help improve clarity. Also, the panel is very small and may benefit from a different presentation in the figure.
We have updated Fig. 1F to include an inset of Quadrant III for improved clarity and readability. We have also moved Fig. S7C to the main text as Fig. 1G and added an expanded explanation for these figures.
(7) Figure 2: The conclusion of a "deeper suppression" of signalling relies on overexpression of tankyrase in an otherwise tankyrase-null background. Have the authors attempted to measure reporter activity or endogenous gene expression without tankyrase overexpression, in Wnt3a-stimulated cells (in the context of a normal Wnt/beta-catenin pathway) or CRC cells at the basal level? Non-catalytic activity in a similar assay has previously been observed upon tankyrase overexpression (10.1016/j.molcel.2016.06.019). Whether or not there is a substantial scaffolding effect at endogenous tankyrase levels after tankyrase inhibition remains unconfirmed, and the PROTAC is a valuable tool to address this important question. The findings presented in Figure S7C and D go some way towards answering this question - these data could be presented more prominently, and similar assays could be performed in other cell systems.
IWR1-POMA suppressed STF activity more effectively than IWR1 in APC-mut DLD-1 and SW480 CRC cells without TNKS overexpression (Fig. S8C). Similarly, IWR1-POMA provided a deeper suppression of STF signals in HeLa cells transfected with AXIN1 and β-catenin while expressing endogenous TNKS (Fig. 4G). These results suggest that inhibitor-induced TNKS scaffolding plays a significant role at endogenous TNKS expression levels. Following the reviewer’s suggestion, Fig. S7C is now Fig. 1G.
(8) Line 237/238: "TNKS accumulation negatively impacts the catalytic activity of the DC (Figure 5D)" - the data do not show this. Beta-catenin levels are a surrogate readout for DC function (phosphorylation and ubiquitylation). Minimally, this requires rewording, with reference to beta-catenin levels.
We have rephrased "TNKS accumulation negatively impacts the catalytic activity of the DC" as "TNKS accumulation negatively impacts the exchange of β-catenin in the DC."
(9) Line 303-304: Beta-catenin is thought to exchange at beta-catenin degradasomes; this is clear from previous FRAP assays and the observation that phospho-beta-catenin accumulates in degradasomes upon proteasome inhibition (10.1158/1541-7786.MCR-15-0125). However, degradasome size hasn't, to my knowledge, been related to activity. Can this be clarified, please?
We apologize for confusing β-catenin phosphorylation with β-catenin abundance. Here, we refer the catalytic activity of the DC to as the ability of the DC to promote β-catenin degradation rather than the kinetics of β-catenin phosphorylation. It is commonly observed that AXIN stabilization by TNKS inhibitors increases the DC size and reduces the β-catenin levels. As such, the induction of AXIN puncta by TNKS inhibitors is frequently used as an indicator of WNT/β-catenin pathway inhibition. However, we have found that, TNKS inhibition drives TNKS accumulation, which reduces the ability of the DC to promote β-catenin degradation. We agree that the DC only primes β-catenin but does not catalyze its degradation. We have revised our manuscript as follows: "increasing the local concentration of the DC components improves its 'effective activity'[50,51]."
(10) There are previous hypotheses/proposals that the sensitivity of CRC cells to tankyrase inhibition correlates with APC truncation or PIK3CA status (10.1158/1535-7163.MCT-16-0578; 10.1038/s41416-023-02484-8). Have the authors considered expanding their cell line panel (Figure S7) to sample a wider range of cell lines, including some that are wild-type with regard to APC or Wnt/beta-catenin signalling in general? This would be a valuable addition to the work. Quantitated colony formation data could be moved to the main body of the manuscript.
We have so far tested the effects of IWR1-POMA on the proliferation of DLD-1, SW480, HT-29, HCT116, and RKO cells (Fig. 6A and 6B). While a heterozygous Ser45 deletion in CTNNB1 confers resistance to IWR1-POMA, we did not observe sensitivity associated with APC or PIK3CA status. The ability of IWR1-POMA to suppress the growth of RKO cells expressing wild-type APC is consistent with a previous report that knockdown of both TNKS1 and TNKS2 stabilized PTEN to suppress cell proliferation and glycolysis in vitro and tumor growth in vivo (PMID: 25547115, Ref. 48) independently of the β-catenin pathway. We have added this new information as well as quantification of the colony growth results (Fig. S8A, S8B, S9A, S9F, and S9G) to the revised manuscript.
(11) The manuscript only mentions toxicity (i.e., therapeutic window) in the last sentence of the Discussion section. As this is THE main challenge with tankyrase inhibitors (as mentioned above), can the authors expand their discussion of this aspect? Is there an expectation that PROTACs may be less toxic?
As discussed above, evidence for on-target toxicity of WNT/β-catenin inhibition is mixed. Yet, the absence of dose-limiting toxicity for basroparib at doses up to 360 mg QD in human (PMID: 40964966, Ref. 64) is encouraging. PROTAC works by catalyzing target degradation, which is different from traditional catalytic inhibitors that require continuous target occupancy at a high level. It remains unclear whether the observed on-target toxicity of TNKSi is associated with TNKS accumulation at high doses, akin to the cytotoxicity induced by PARP1-trapping upon catalytic inhibition. We have included a brief discussion of the toxicity issue in the final paragraph of the Discussion section.
(12) Figures 3, 4, 5A: For fluorescence microscopy experiments, can these be quantified, and can repeat data be included?
We have included quantification data and replicate information for Fig. 3–5.
(13) Figure 4, S6: An additional channel illustrating the distribution of cells (e.g., nuclei, cytoskeleton, or membrane) would be helpful for orientation and context for the AXIN1 signal.
We have included cell outlines or nuclear staining for Fig. 3, 4, S6, and S7.
(14) How were cytosolic fractions of cells prepared to assess cytosolic beta-catenin levels? This detail is missing from the methods.
We have updated the Methods section to include additional details on the preparation of the cytosolic fractions of cells.
Reviewer #3 (Public review):
In this manuscript, Wang et al employ a chemical biology approach to investigate the differences between the enzymatic and scaffolding roles of tankyrase during Wnt β-catenin signalling. It was previously established that, in addition to its enzymatic activity, tankyrase 1/2 also plays a scaffolding function within the destruction complex, a property conferred by SAM-domain-dependent polymerization (PMID: 27494558). It is also known that TNKS1/2 is an autoregulated protein and that its enzymatic inhibition leads to accumulation of total TNKS proteins and stabilization of Axin punctae (through the scaffolding function of TNKS1/2), leading to rigidification of the DC and decreased β-catenin turnover. The authors surmised that this could, in part, explain the limited efficacy of TNKS1/2 catalytic inhibition for the treatment of colorectal cancers. To test this hypothesis, they evaluated a series of PROTAC molecules promoting the degradation of TNKS1/2 to block both the catalytic and scaffolding activities. They show that IWR1-POMA (their most active molecule) promotes more efficient suppression of beta-catenin-mediated transcription and is more active in inhibiting colorectal cancer cell and CRC patient-derived organoids growth. Mechanistically, the authors used FRAP to demonstrate that catalytic inhibitors of TNKS led to a reduced dynamic assembly of the DC (rigidification), whereas IWR1-POMA did not affect the dynamics.
Overall, this is an interesting study describing the design and development of a PROTAC for TNKS1/2 that could have increased efficacy where catalytic inhibitors have displayed limited activity. Knowing the importance of the scaffolding role of TNKS1/2 within the destruction complex, targeting both the catalytic and scaffolding roles certainly makes sense. The manuscript contains convincing evidence of the different mechanisms of the PROTAC vs catalytic inhibitors. Some additional efforts to quantify several of the experiments and to indicate the reproducibility and statistical analysis would strengthen the manuscript. Ultimately, it would have been great to evaluate the in vivo efficacy of IWR1-POMA in an in vivo CRC assay (APCmin mice or using PDX models); however, I realize that this is likely beyond the scope of this manuscript.
We thank the Reviewer for the helpful suggestions.
I have some recommendations listed below for consideration by the authors to strengthen their study:
(1) The title is slightly misleading, as it is already known that the scaffolding function of TNKS is important within the DC. The authors should consider incorporating the PROTAC targeting aspect in the title (e.g., PROTAC-mediated targeting of tankyrase leads to increased inhibition of betacat signaling and CRC growth inhibition).
We have modified the title accordingly to "Targeting tankyrase scaffolding in the β-catenin destruction complex by PROTAC overcomes the limitation of catalytic inhibitors in cancer."
(2) The authors should comment in the manuscript on the bell-shaped curve obtained with treatment of cells with the PROTACs (Figure S2C). This likely indicates tittering of the targets within a bifunctional molecule with increasing concentration (and likely reveals the auto-inhibition conferred by the catalytic inhibition alone).
As suggested by the Reviewer, the bell-shaped dose-response likely originated from the formation of non-productive binary protein-ligand complexes at high PROTAC concentrations. We have added a sentence to clarify this unique behavior of PROTAC molecules.
(3) The authors comment that using G007-LK as warehead was unsuccessful, but they do not show data. Do the authors know why this was the case?
The structure-activity relationship of PROTACs is often unpredictable, as both the kinetics and thermodynamics of target and E3 ligase binding play important roles in promoting efficient target degradation. We have include data on G007-LK based PROTACs (Fig. S2D) in the revised manuscript.
(4) Throughout the manuscript, the authors need to do a better job at quantifying their results (i.e., the western blots and the IF). For example, the degradation of TNKS1/2 in Figure 1D is not overly convincing. Similarly, the IF data in Figure 3 needs to be quantified in some ways. Along the same lines, the effect of IWR1-POMA treatments on the proliferation of cells and organoids should be quantified using viability assays... There is also no indication of how many times these experiments were performed and whether the blots shown are representative experiments. The quantification should include all experiments.
We have included quantification of the immunofluorescence images, colony formation data, and Western blots in the revised manuscript.
Recommendations for the authors:
Reviewer #2 (Recommendations for the authors):
(1) For clarity, can the authors use the official gene names, TNKS and TNKS2?
We favor using TNKS1 and TNKS2 when referring to the protein for clarity and use TNKS for simplicity when referring to both proteins.
(2) Line 92: The authors refer to TNKS2 "induction" - it remains unclear what is meant by "induction".
We have changed "without induction" to "under basal conditions".
(3) Can the authors please display molecular weight markers for Western blots throughout?
(4) Line 144: The description "significantly more effectively" refers to Figure S5A, which shows a single, non-quantified Western blot. I don't think significance has been tested, and this statement should be reworded, or quantified aggregate data provided.
We have added a Supplementary Information file showing molecular weight markers and quantification of Western blots.
(5) Line 226: "plateaued at a much lower level" - can this be expressed more quantitatively in the text?
We have included more quantitative information on the FRAP results.
(6) Line 249: Can the authors repeat the cross-reference to Figure S7A here?
We have repeated the cross-reference to the figures.
(7) Line 266: The description of the experiment using the GSPT1/2 degrader CC-90009 would benefit from a brief recap of the purpose as not every reader will be familiar with this common PROTAC off-target. This is a very thorough analysis, though, and commendable.
We have added background information on GSPT1 degradation to the revised manuscript.
(8) Figure 1A: Can the number of repeats and the type of repeats be indicated, please?
(9) Figure 2: Does n refer to biological or technical repeats?
(10) Figure 5B, D: How many separate experiments are the data based on?
(12) Figure S3D, S9A, D: number and types of repeats and the nature of the displayed data and error bars need to be included, please.
(13) Figure S6B, S7B: I can see three data points, but it would still be helpful to state the number and type of repeats in the legend.
(14) Figures S9A, S9D: There is value in showing the cumulative data from several repeats in the main figure (Figure 6, which currently is only qualitative) rather than the supplementary material.
(15) Where single Western blots are shown, can the authors indicate how many experiments they are representative of?
We have included the number of biological repeats for all data.
(11) Figure S2C: For most graphs, the main response of interest occurs at low compound concentrations. The y-axis scale does not always help the reader to appreciate the effects, as the response seems small against the magnitude of the hook effect. Interrupting the y-axis as in the final panel may help, with y-axis scales consistent over all panels in the figure.
We have updated Fig. S2C to emphasize on the degradation efficacy.
(16) The authors may want to give further method details for some of their assays to facilitate replication of their experiments in the future. For example, the STF assay description is currently quite minimalistic. I assume the assay is fairly robust, though. Other details include cell media (general media details and specific additives and their concentrations in the 3D spheroid formation assay), etc. A general look at the methods section will likely be beneficial.
We have updated the Methods section to provide more detailed experimental information.
Reviewer #3 (Recommendations for the authors):
(1) In Figure 2A, one of the most important findings of the manuscript is that IWR1-POMA induced promoted deeper suppression of beta-catenin-mediated transcription. This seems to be the case only at 3.2uM. Is it statistically significant? What are the data points on this graph? What are the error bars?
We have included statistical analysis as Fig. S5G.
(2) On Figure 2C and 2D, do the authors know why the TNKS20M1054V mutant is much better at promoting signaling than the TNKS1-PD ? Is it expression levels?
It is indeed interesting that TNKS2-M1054V promoted significantly stronger WNT signaling than TNKS1-PD. The basis for its strong scaffolding effect is unclear.
(3) In Figure 4C, the authors claim that when cells are treated with IWR1-POMA, AXIN1 is distributed diffusely throughout the cytoplasm. It appears that small punctae are visible.
Quantitative analysis (Fig. 4F) suggest that the size of AXIN1 puncta upon IWR1-POMA is rather insignificant.
(4) Label on Figure 1D has a spelling error TNKS1/2.
Corrected.
