Protein stabilization of ITF2 by NF-κB prevents colitis-associated cancer development

Chronic colonic inflammation is a feature of cancer and is strongly associated with tumorigenesis, but its underlying molecular mechanisms remain poorly understood. Inflammatory conditions increased ITF2 and p65 expression both ex vivo and in vivo, and ITF2 and p65 showed positive correlations. p65 overexpression stabilized ITF2 protein levels by interfering with the binding of Parkin to ITF2. More specifically, the C-terminus of p65 binds to the N-terminus of ITF2 and inhibits ubiquitination, thereby promoting ITF2 stabilization. Parkin acts as a E3 ubiquitin ligase for ITF2 ubiquitination. Intestinal epithelial-specific deletion of ITF2 facilitated nuclear translocation of p65 and thus increased colitis-associated cancer tumorigenesis, which was mediated by Azoxymethane/Dextran sulfate sodium or dextran sulfate sodium. Upregulated ITF2 expression was lost in carcinoma tissues of colitis-associated cancer patients, whereas p65 expression much more increased in both dysplastic and carcinoma regions. Therefore, these findings indicate a critical role for ITF2 in the repression of colitis-associated cancer progression and ITF2 would be an attractive target against inflammatory diseases including colitis-associated cancer.

1) Firstly, while ITF2 is clearly relevant to CAC inhibition in the AOM/DSS and DSS models, the clinical significance of the proposed mechanism in the control of CAC progression in CD/UC patients is not clear. It is also unclear from the limited nuclear localization data presented whether ITF2 is a key regulator of NF-κB activity in these clinical settings and whether and how Parkin is involved in the regulation of ITF2 stability in human CD/UC. For instance, is there a correlation between ITF2 (or Parkin) expression and disease severity/clinical outcomes in UC/CD or CAC patients? Further, is there any evidence of ITF2 loss-of-function or Parkin gain-of-function mutations occurring in human CAC? Are there any SNPs in these genes or their regulators that might predispose UC/CD patients to CAC? Crucially, what downregulates ITF2 expression during the transition from dysplasia to carcinoma in CAC patients? Any data addressing at least some of these points would considerably strengthen the clinical relevance of the proposed mechanism.
2) Secondly, it is unclear that the observed effects of ITF2 in AOM/DSS-driven CAC are directly related to Parkin-dependent regulation or NF-κB inhibition. a) The ITF2 interactions with Parkin and RELA/p65 were investigated in overexpression systems, and there is no evidence that endogenous proteins engage in these interactions either in in vitro cell culture systems or intestinal epithelial cells in vivo, which could be examined for instance by using PLA assays. b) Furthermore, the molecular interactions of ITF2 with RELA/p65 and Parkin are only superficially characterized, thus precluding the generation mutant proteins with reduced binding affinity, which could provide useful tools to establish whether the biological effects of ITF2 on CAC suppression are indeed mediated by NF-κB inhibition. Similarly, it would be helpful to determine whether Parkin conditional deletion in intestinal epithelial cells has opposite phenotype of ITF2 deletion in AOM/DSS-dependent CAC and whether the effects of Parkin loss in this model are reversed by ITF2 deletion. Any data along these lines would significantly strengthen the manuscript conclusions and the proposed mechanism of action of ITF2 in suppressing CAC pathogenesis.
3) Thirdly, there are other technical limitations. For instance, many key experiments were conducted using a single CRC cell line, raising the possibility that the observed biological effects could have resulted from idiosyncrasies of that particular cell line. Moreover, many of the inhibitors used in Fig. 1 are not specific for their declared targets. 4) Lastly, the text presents some apparent inconsistencies and issues that have not been adequately discussed or explained. For example, it is stated that previous studies have shown increased ITF2 mRNA expression in UC and CD patients relative to normal individuals, with a positive correlation with TNA levels (refs 22-24). Yet, the authors' own data show no changes in ITF2 mRNA levels between healthy individuals and UC/CD patients (Fig. 2I). Further, no changes in ITF2 mRNA expression were observed in CRC cell lines following TNF stimulation ( Fig. 2A-2B).
Reviewer #2 (Remarks to the Author): This is an interesting manuscript describing the role of Immunoglobulin transcription factor 2 (ITF2) in colonic inflammation and colitis associated tumorigenesis. ITF2, also known as TCF4, is a bHLH transcription factor that targets E boxes and prior studies suggest that it is a context dependent tumor suppressor or oncogene. ITF2 is upregulated in direct relation to TNF in IBD. P65 was also upregulated under inflammation and was demonstrated to stabilize ITF2 protein by disrupting Parkin (an E3 ubiquitin ligase):ITF2 interactions. The ITF2:p65 interaction domain mapped to N-terminus of ITF2 and carboxy tail of p65, inhibiting ubiquitination. They then demonstrate that epithelial deletion of ITF2 resulted in increased nuclear p65 and increased AOM/DSS induced tumorigenesis. ITF2, while upregulated in IBD, is lost in human CAC samples, whereas p65 was increased. Overall, these are well designed and executed experiments and an important discovery. I do have a few major and minor critiques.
Major critiques: 1) The microbial differences can have a large impact on DSS and AOM/DSS modeling. What measures were taken to control for microbial differences between experimental groups? These should be clearly stated in the methods.
2) The methods section indicates that half the colons were excised for RNA and protein analysis, and half for histopathologic analysis. DSS commonly has a larger impact on the distal colon, in fact that appears to be supported by looking at the tumor distribution in the AOM/DSS experiments (predominantly distal tumors). The authors should describe which "halves" were obtained. Proximal vs distal half or where the colon halved longitudinally? Also, if proximal and distal halfs, how does this confound their work? 3) I am not sure that I completely follow the model picture. It seems that the last panel should be in the setting of cancer, not inflammation, demonstrating that loss of ITF2 results in increased nuclear p65 and upregulated target genes. 4) The authors convincingly demonstrate that ITF2 is upregulated with DSS treatment. Did they observe reductions with AOM/DSS in the tumors and what were it's levels in the adjacent "normal". I would predict reduction in tumors coincident with p65 nuclear localization and presence in the adjacent normal. It would be interesting to know b/c if ITF2 was also lost in the "adjacent normal" it might suggest a field or priming effect with ITF2 loss. 5) Can the authors state if the organoid data is colon or small intestine. Treatment of colonoids is more relevant to the current report.
Minor: 1) Thorough proofing of the manuscript for typographical and grammar/usage errors should be performed.
Reviewer #3 (Remarks to the Author): In this manuscript, Lee and coauthors identified that ITF2 (also known as TCF4) binds p65/RelA, a NF-κB transcription factor, and that stabilizes ITF2. Furthermore, the authors showed that parkin is a ubiquitin ligase (E3) to induce proteasomal degradation of ITF2. Then, the authors concluded that the ITF2-p65-parkin axis plays an important role in colitis-associated cancer. Although interesting findings have been made in this study, the cellular and pathological mechanisms have not been fully elucidated.
Major comments: 1. In Fig. 1, the authors showed that ITF2 expression is positively correlated with p65/RelA levels.
Among five NF-κB factors, RelB and c-Rel, but not p50or p52, have similar domain structure with that of p65. Do RelB and c-Rel affect ITF2 binding, stabilization, and ubiquitination like p65? Please show the expression levels of p65 in Fig. 1C. In Fig. 1D, the pretreatment with TPCA1, an IKK inhibitor, seems to suppress TNF-α-induced IKK activation. Then, why are p65 and IκBα were degraded without NF-κB activation? In Fig. 1C and 1D, the bands visible on the border below the ITF2 blot appears to be augmented in the presence of Bay 11-7082 and TPCA1. It is necessary to submit uncropped immunoblotting images to clarify whether they are dephosphorylated ITF2 by the inhibition of IKK. 2. In Fig. 2E and 2F, the authors showed that cells treated with TNF exhibited a higher half-life of ITF2 than that without TNF-α. Upon TNF-α stimulation, p65 basically translocate into the nucleus and function as a transcription factor with shuttling to cytosol. Thus, it is unclear where the physiological p65-ITF2 binding occurs to stabilize ITF2. Moreover, in the presence of TNF-α and CHX, cells induce RIP1-independent apoptosis. Does ITF2 regulate TNF-α-mediated cell death pathways, such as apoptosis and necroptosis? 3. In Fig. 3, although the authors showed the interaction of p65 and ITF2 by overexpression system, the authors need to analyze the endogenous interaction of ITF2 and p65 with or without TNF-α stimulation. 4. In Fig. 4A and on lines 443-444, the authors described that parkin is the only E3 among ITF2inteacting proteins in the BioGIRD software. However, when I checked the BioGIRD database, TCF4 (ITF2) has 281 unique interactors (https://thebiogrid.org/112787/summary/homosapiens/tcf4.html), and besides PARK2, E3s such as CUL4B, IRF2BP2, RNF14, RNF6, TRIM24, TRIM68, UHRF2, and so on, are listed. Is parkin really the only E3 for ITF2? Basically, parkin is a latent E3, which is activated by PINK1-meidated phosphorylation of the parkin-UBL domain and/or ubiquitin. Is this kind of phosphorylation happening? Does mitochondrial damage that physiologically activates parkin affect the proteasomal degradation of ITF2? The authors need to investigate the parkin-induced ubiquitination site(s) in ITF2 and linkage of polyubiquitin chain. In vitro ubiquitination assay, as shown in Fig. 4D, is very weak. HeLa cells do not express parkin, but does ubiquitination occur when ITF2 and HA-Ub were co-expressed in HeLa cells? Moreover, to clarify the contribution of parkin in CRC, please show the expression levels of parkin in cells shown in Fig. S2A. 5. In Fig. 5C, the authors estimate the mechanism by which ITF2 suppresses the nuclear translocation of p65 by binding to p65, but this is an unstimulated state and cannot be evaluated. It is necessary to analyze the dynamics of p65 associated with inflammatory cytokine stimulation as mentioned above. 6. In Fig. 6, show p65 and parkin expression with IB and IHC. Differences in NF-κB target genes expression and cell death induction in tumor and non-tumor areas also need to be analyzed.

Statement of Revision to Reviewer #1
We would like to begin by expressing our appreciation for the reviewer's thoughtful and considerate suggestions and comments. we hope our responses will be deemed satisfactory.
Thank you for the reviewer's critical reading and suggestions for strengthening our conclusions. In Supplementary Figure S15 Regarding the next question, although this issue is a crucial point of the manuscript and has to be further addressed, we cannot find appropriate publications about what the reviewer exactly mentioned. We do believe that there are, to our knowledge, any related papers showing evidence of ITF2 loss-of-function or regulators of ITF2. However, as shown in Figure 7 and Supplementary figure S15, we revealed that ITF2 protein expression levels are highly lost during the dysplastic-to-carcinoma transition. Especially, the patients who experienced severe CAC clinical symptoms (stages 3 and 4) presented much lower ITF2 levels compared to the early stages (stages 0-2). Consistently, the microarray dataset (GSE 3629, shown in Figure 7C) displayed rare ITF2 mRNA expressions in UC-associated neoplastic legions, UC-associated carcinoma tissues, and sporadic CRC tissues, suggesting that the levels of ITF2 have already disappeared in the mRNA levels. We therefore highly suspect that ITF2 expression is genetically lost during the dysplasia-to-carcinoma transition in CAC patients as sporadic CRC patients did. 1,2 For future works, we'll scrutinize whether ITF2 has LOF or GOF in CAC patients, and also it'll be interesting to find possible ITF2 regulators in the dysplastic-tocarcinoma transition. As the reviewer knows, it requires a huge time for recruiting appropriate patients (especially, CAC patients) and gathering related patient information. We understand the reviewer's concerns but would ask that the reviewer show some leniency on this point.
Because it is still an important issue, we later will do using a well-designed plan. We hope that this answer would satisfy your question.

1.
Shin In Supplementary Figure S10 Regarding the second question, we decided to identify the Parkin binding sites to ITF2 to see the molecular interactions deeply. We carried out the in vitro binding assay and confirmed that ITF2 and Parkin bind together ( Figure 4C). Immunoprecipitation assay showed that Parkin binds to the N-terminals of the ITF2 ( Figure 4D). We then utilized one of the 3D computational protein binding prediction programs (AlphaFold 2) to characterize promising amino acids to which Parkin can bind to the ITF2. AlphaFold2 results revealed that Parkin binds to the Nterminals of the ITF2 between the lysine residues (171-175) ( Figure 4E). After generating the ITF2 mutant which is deleted in 5-amino acids (Lys171-175), we then transfected the mutant ITF2 into the HeLa cells which have been reported to be little or no endogenous Parkin expression 1,2 in the absence or presence of Parkin to determine whether these lysine residues are the major ubiquitination sites of the ITF2 by the Parkin. As expected, Pakin induced ITF2 ubiquitination, but the levels seem to be downregulated when the cells are transfected with ITF2 mutant (∆K171-175), suggesting that the lysine residues between the 171-175 of the ITF2 are the major ubiquitination sites by the Parkin (Figure 4F and Supplementary Figure S9D).
We included those results in Figure 4 and Supplementary Figure S9, and modified the manuscript as follows; Manuscript Page 9, Line 199,

….Protein-protein binding sites prediction
Alphafold2 predicts three-dimensional (3D) protein structures based on the protein sequences of the ITF2 and Parkin, and then showed promising Lys residues that are supposed to be  Figure S9D).
Co-immunoprecipitation and in vitro binding assay confirmed that Parkin can bind to ITF2 ( Figures 4B, C), and it turns out to be that Parkin binds to the N-terminals of ITF2B ( Figure   4D). To corroborate which sites of ITF2 are associated with Parkin-induced ITF2 ubiquitination, we generated an ITF2 mutant where key lysine residues (Lys171-175) are removed based on the AlphaFold2 results ( Figure 4E). As expected, when the HeLa cells are overexpressed with Parkin and WT ITF2, we observed upregulation of ITF2 ubiquitination.
In Figure 4 In Supplementary Figure S9 For the final question, as the reviewer knows, Parkin has been reported to activate NF-B activity in multiple different tissues regardless of ITF2 expression 3,4,5 , so we could expect reduced inflammation, diminished tumor numbers, or possibly, decreased tumor sizes in the Parkin cKO mice. However, even though Parkin cKO mice showed the opposite results compared to the results presented in ITF2 cKO mice, we might think that it is hard to say Parkin deletion reduced NF-kB activity through the ITF2 stabilization method. As we discussed in question 1, we introduced Parkin as a novel E3 ligase for ITF2 ubiquitination, but we highly suspect that ITF2 seems to be genetically lost during dysplasia-to-carcinoma transition rather than Parkin-dependent mechanisms (Figure 7). In addition, when we take a look at the Parkin expression patterns in the CAC patient tissues, there is no significant difference between dysplasia and carcinoma regions (Supplementary Figure S15C). We understand the reviewer's concern and it would have been better if we could generate double knock-out mice (e.g. generate Parkin cKO in the ITF2cKO mice) and check whether those results could happen.
Indeed, we were looking into the Parkin floxed mice to generate intestinal-specific parkin deletion mice but, unfortunately, it was not commercially available and the timeline was not allowed us to get or generate a new mouse strain. Besides, the COVID period makes all things much slower. Hopefully, this answer would satisfy your question. of the experiments were done on at least two different CRC cell lines or carried out in an overexpression system to reveal underlying mechanisms. We agree with the reviewer's opinion that many of the inhibitors described in Figure 1C are not that specific for their declared target but rather affect multiple targets. For example, Rapamycin majorly inhibits the mTOR signaling pathway but also could affect multiple targets which were responsible for cell growth and cell cycle progression. In the same context, AG490 partially affects the NF-B pathway as well as blocks the JAK-STAT pathway. However, those inhibitors mentioned in the manuscript are highly selective for the described target (the declared target means major target, not means single target), thus various researchers used those inhibitors for the same purposes we do believe. Most importantly, the NF-B inhibitor (Bay 11-7082) shown in Figure 1C was the only chemical that could attenuate TNF-induced ITF2 upregulation even though the other inhibitors might affect multiple signaling pathways. As mentioned above, AG490 reportedly could partially inhibit the NF-B activation and that is why AG490 pre-incubation presented a half reduction of the TNF-induced ITF2 expressions. To reduce the reviewer's concern that the observed biological effects could have resulted from idiosyncrasies of that particular cell line, we looked into the other kinds of NF-B inhibitors such as TPCA1 and Bay-11-7085 (reportedely, much more specific for NF-B) and then performed similar experiments in two additional CRC cell lines (SW480 and WiDr) including Colo320DM cells ( Figure 1D and Supplementary Figure S3). TPCA1 and Bay-11-7085 have been known to be potent IKK inhibitors that enable to block of TNF--induced phosphorylation of IkB-, resulting in an inhibition of NF-B. As expected, pre-incubation of the TPCA1 or Bay-11-7085 lessened TNF-induced ITF2 upregulation as Bay-11-7082 did, suggesting that NF-kB is involved in the TNF-induced ITF2 expression. In addition, p65 knockdown and overexpression studies presented in Figure 1-4 confirmed that p65 is responsible for TNF-induced ITF2 expression.
As reviewer 3 requested p65 immunoblot and uncropped immunoblot of ITF2 in Figure 1C (Question 1) to examine whether the bands visible on the border below of the ITF2 blot related to dephosphorylation issue, we reconducted this experiment using the same samples to check whether this issue is a technical problem or dephosphorylation. To avoid any confusion, we modified the previous figures to the current ones.
We added these data to Figures (Fig. 2I). Further, no changes in ITF2 mRNA expression were observed in CRC cell lines following TNF stimulation ( Fig. 2A-2B).
Answer: We apologize for this confusion and we agree that these inconsistencies have to be addressed in the discussion part. As the reviewer's pointed out, previous reports (refs [22][23][24] showed increased ITF2 mRNA expression in both UC and CD patients compared to healthy individuals whereas our data ( Fig. 2A, B) and the data from GEO databases (Fig. 2I) revealed that the levels of ITF2 are comparable across the samples.
If we take a guess, we might think that the reason is coming from the inconsistent sample collection method from the patient biopsies. The reference papers (refs 22-24) and most of the GEO datasets mentioned that they used biopsies from the colonic mucosa derived from healthy, UC, or CD which enables possible contamination of lymphocytes. It is unclear how well the sample collection was controlled. For example, some of the UC or CD samples were taken from relatively normal regions but sometimes, the samples could contain relatively high inflammatory locations. In addition, there is a possibility that the authors did epithelial cell sorting before running the bulk RNA-seq. As the reviewer knows, ITF2 reportedly binds to the immunoglobulin enhancer Mu-E5/KE5-motif to facilitate immunoglobulin expression. Thus, if the RNAs were extracted from the colonic mucosa, and not even sorted, mucosa samples could randomly contain lymphocytes, making it hard to interpret the data. That is why many of the sequencing data showed contradictory results. However, most importantly, our in-vitro data (Figure 2A, B) and GSE 11223 dataset which was done in epithelial biopsy showed that the ITF2 mRNA levels were comparable across the groups.
We discussed this issue in the discussion part as follows;  Figure 2I exhibited no significant changes in ITF2 mRNA levels across the samples. We might suspect that this conflict result would be coming from the inconsistent sample collection method from the patient tissues. The papers mentioned above mostly utilized biopsies from the colonic mucosa derived from healthy, UC, or CD patients which enable possible contamination of lymphocytes. Besides, there is a chance that samples might be collected in the randomly selected areas, which means the ratio of the lymphocytes and epithelial compartment could be arbitrarily decided. Thus, if the RNAs were extracted from the colonic mucosa, and not even sorted, mucosa samples could randomly contain lymphocytes, making it hard to interpret the data. That is why many of the sequencing data showed contradictory results. However, most importantly, our in-vitro data (Figure 2A, B) and GSE 11223 dataset which was done in epithelial biopsies showed that the ITF2 mRNA levels were comparable….

Statement of Revision to Reviewer #2
We authors thank the reviewer for the comments, which contributed to the improvement of this manuscript.

Major comments Question 1. The microbial differences can have a large impact on DSS and AOM/DSS
modeling. What measures were taken to control for microbial differences between experimental groups? These should be clearly stated in the methods.
Answer: We agree with your opinion that microbial differences could highly impact DSS and AOM/DSS mouse models. One report showed that Germ-Free (GF) mice developed significantly more and larger tumors compared with that Specific-Pathogen-Free (SPF) mice after AOM and DSS treatment. 1 In addition, two different papers also presented that both IL-2-deficient and IL-10-deficient mice, which under conventional conditions develop spontaneous colitis, have significantly reduced or absent intestinal inflammation in germ-free conditions. 2,3 To sustain a similar condition, all the mice treated with DSS and/or AOM/DSS were bred and maintained in SPF conditions. SPF mice were fed the same autoclaved chow diet and used the same drinking water offered by the SPF room. As all animal facilities do, our animal facility routinely (mostly at 6w intervals) runs PCR, ELISA, or culture to monitor possible bacterial & mycoplasma contamination. Therefore, we do believe that our experiments were done at least under similar conditions. We already described some information in the method section of the main manuscript, but we also included this information as follows; Manuscript Answer : We are sorry about the confusion. As mentioned in the manuscript, we excised the colon horizontally (proximal vs distal), and then took the distal colons for histology ( Figures   1, 6 and Supplementary Figures S4, S11, S12, and S13) and used the other half for protein work (western blot, only for Figure 6B). Figure 6B was presented to show whether ITF2 is deleted in the colonic epithelial cells specific manner. We, therefore, might think that the tissues coming from the other half (proximal part) were not going to be a problem as ITF2 knockdown was evaluated in the same proximal part of the tissues. We also took some portion of the proximal tissues and then extracted the RNA, but have not used it in all of our experiments. We just stored the RNA samples just in case. However, based on the reviewer's opinion, we should have divided the tissues as longitudinally. We'll reference it for future work.
To avoid any confusion, we deleted the information on how we can get RNA from the colon as follows; Manuscript Answer: We appreciate your kind opinion. Based on the reviewer's suggestion, we changed the wording "inflammatory condition" to "colitis-associated cancers (CAC) with ITF2 (-).
We modified the graphical summary as follows; In Figure 7D,

Question 5. Can the authors state if the organoid data is colon or small intestine.
Treatment of colonoids is more relevant to the current report.

Answer:
We apologize for providing insufficient information. We performed the organoid culture using crypt epithelial cells from the mouse colon tissues. Based on the reviewer's suggestion, we changed the wording "IEC organoids" to "colonic organoids" for making sure where they were coming from.
We revised the manuscript as follows; Answer: Based on the reviewer's suggestion, we revised the manuscript to change typographical and grammatical errors.

Statement of Revision to Reviewer #3
We thank the reviewer for raising these issues and hope that our response will be satisfactory.
Major comments Question 1. In Fig. 1 Answer: We are thankful for the reviewer's kind comments. Following the reviewer's suggestion, we first examined whether RelB and c-Rel could bind to the ITF2, and then evaluated whether these proteins were also associated with the stabilization and ubiquitination of ITF2 as p65 did. Although RelB and c-Rel can bind to the ITF2 (Supplementary Figure   S8A)., they cannot affect the expression and ubiquitination levels of the ITF2 ( Supplementary   Figures S8B, C). Taken together, among the three different proteins that have similar domain structures, we suggest that p65 is the only protein that can affect ITF2 expression levels by regulating ubiquitination.
We included those results in Supplementary Figure S8, and modified the manuscript as follows; Manuscript Page 20, Line 459,  Figure S8,

In Supplementary
We also included the p65 immunoblot in Figure 1C as the reviewer requested. As shown in Figure 1D, when the cells were pretreated with NF-B inhibitor (Bay 11-7082), the p65 expression level was downregulated. To confirm this issue, we included additional NF-B inhibitors such as TPCA1, Bay 11-7085, and then examined those effects on p65 in three different CRC cell lines (Colo320, SW 480, WiDr). We observed very similar results in each cell line ( Figure 1D, Supplementary Figure S3) We included these figures and modified the manuscript as follows; In Figures  Regarding the next question, p65 and IkB- seem to be degraded by the BAY 11-7082 and Bay 11-7085 as well as TPCA1 even though the samples were treated with TNF (Figures 1 C, D, and Supplementary Figure S3). We truly have zero clues why those inhibitors alleviate p65 and IkB- expression levels, but when we confirmed these issues in three different CRC cell lines, we observed similar phenomena as we showed before ( Figure 1D and Supplementary Figure S3). Although we cannot explain the underlying mechanisms of how those inhibitors affect p65 and IkB- reductions, previous publications also showed IkB- reduction by the TPCA1 or Bay 11-7082 incubation 1,2 . However, p65 seems to be quite stable even after the treatment of inhibitors. We speculate that these might be distinct features in the colonic cells but not occurred in other cell types or there is a chance that p65 has a positive feedback loop to maintain the p65 expression itself. Thus, interrupting p65 activation by the inhibitors might affect p65 expression itself. Hopefully, this answer and our additional experiment would satisfy your question.
For the last question, we re-electrophoresed the samples shown in Figure 1C to resolve whether the bands which were visible on the border below the ITF2 blot in Figure 1C are the dephosphorylated forms of ITF2 or technical errors. So, we decided to elongate the electrophoresed time around 1 h to make more room between the ITF2 band and the band visible on the border below the ITF2 as the two different bands were too close to evaluate something. However, when we ran the samples for a much longer time, "the expected dephosphorylated forms of ITF2" go far down more than 20kDa, and that results led us to think about the technical errors. Besides, NF-B inhibitors used in this study have been known to be representative specific inhibitors for blocking IkB- phosphorylation by inhibiting IKK. As the reviewer knows, IKK is tightly regulated, highly stimulus-specific, and target-specific (which means IkB-) which is essential for a plethora of functions attributed to NF-kB. To avoid any confusion, we changed the current data to a newly modified one which was electrophoresed longer time.
We added those results in Figures  We added this data to Figure 5D and revised the manuscript as follows; Manuscript Page 11, Line 241,

…..Fractionation of cytoplasmic and nuclear components
HCT 116 stable cell lines established with small hairpin RNA targeting ITF2 20 were used and treated with or without TNF (Peprotech, Rocky Hill, CT, USA). The nuclear fraction experiment was performed as previously described. 32 ....
Manuscript Page 22, Line 514, ….We also conducted a similar experiment under TNF-treated conditions to mimic inflammatory conditions. We observed that plenty of p65 was left in the cytosol even after TNF treatment, implying that p65 has a chance for physiological binding with ITF2 in the cytosol under inflammatory conditions. Most importantly, the amount of translocated p65 to the nucleus by the TNF challenge was downregulated by the ITF2 overexpression ( Figure 5D)….
In Figure 5D, Regarding the second question, we transfected the Colo320DM cells with ITF2 followed by the TNF stimulation for 8h or 24h respectively. Annexin V-PI staining results revealed that the proportion of the apoptotic cells and necrotic cells was comparable in the early timepoint (8 h), but apoptotic cells were reduced at a half ratio when they were incubated with TNF for 24 h.
As the necrotic cell population seems to be comparable, we didn't measure the necroptosis cell population in the same condition. We might conjecture that 8h stimulation of TNF is too short to make any changes (the timepoint used in our study).
We included this data in Supplementary Figure S5 and revised the manuscript as follows; Supplementary Manuscript Page 5, Line 116,

…FACS analysis of cell death
The effect of ITF2 on TNF-mediated cell death pathways such as apoptosis and necrosis was examined by the combined application of Annexin V-fluorescein isothiocyanate (FITC) apoptosis detection kits (BD Pharmingen, Heidelberg, Germany) and propidium iodide In Supplementary Figure S5 Question 3. In Fig. 3, although the authors showed the interaction of p65 and ITF2 by overexpression system, the authors need to analyze the endogenous interaction of ITF2 and p65 with or without TNF-α stimulation.
Answer: As the reviewer commented, we performed an immunoprecipitation analysis to see the interaction of p65 and ITF2 in endogenous conditions with or without TNF stimulation.
Consistently, they were bound together and the binding amount was increased by the TNF challenge.
We added these results in Supplementary Figure S10, and modified the manuscript as follows; Manuscript Page 21, Line 493, …Consistently, co-immunoprecipitation assay against endogenous proteins as well as PLA assay indicated that ITF2 interacts with p65 and Parkin ( Supplementary Figures S10A-C). The endogenous interaction between p65 and ITF2 was notably increased following the TNF stimulation (Supplementary Figure S10D) Answer: We appreciate the reviewer for the important comment. Among the E3s that the reviewer mentioned, we firstly screened the E3 proteins based on the experimental evidence which was done by the two-hybrid system rather than the other methods, and then we finally got 3 proteins (PARK2, RNF6, and TRIM68) ( Figure 4A) Figure S9D).
To determine which sites are crucial for Parkin-mediated ITF2 ubiquitylation, we conducted an immunoprecipitation experiment and observed that Parkin binds to the N-terminals of the ITF2 (Figures 4C, D). We then exploited one of the 3D computational protein binding prediction programs (AlphaFold2) to identify promising lysine residues to which parkin could bind to the ITF2. AlphaFold2 results presented that the lysine residues (Lys171-175) of the ITF2 seem to be promising ( Figure 4E). We, therefore, generated the ITF2 mutant (∆K171-175) and then transfected it into the HeLa cells in the absence or presence of Parkin.
Importantly, the levels of ubiquitination by the Parkin are downregulated when the cells are transfected with mutant ITF2, suggesting that the lysine residues between the 171-175 of the ITF2 are the major ubiquitination sites by the Parkin ( Figure 4F). As we investigated the parkin-induced ubiquitination sites in ITF2 which is more important than the in-vitro binding of ITF2 and Parkin, we moved in-vitro binding data to Supplementary Figure S9E.
Based on the experimental results, we added those results in Figure 4 and Supplementary Figure S9, and revised the manuscript as follows; Manuscript Page 9, Line 199

..Protein-protein binding sites prediction
Alphafold2 predicts three-dimensional (3D) protein structures based on the protein sequences of the ITF2 and Parkin, and then showed promising Lys residues that are supposed to be ubiquitinated by the Parkin in the N-terminus of the ITF2. 30 Manuscript Page 15, Line 342 ..Site-specific deletion mutation of ITF2 was performed using PCR-based mutagenesis….
Manuscript Page 20, Line 470 …Among the 284 unique ITF2-interacting proteins displayed in the BioGRID, we found 9 candidates including PARK2 which have an E3 ubiquitin-protein ligase role ( Figure 4A), and then we narrowed down to the 3 promising targets based on the experimental evidence which was demonstrated by the two-hybrid system. However, the knockdown experiment presented that only Parkin was responsible for the ITF2 ubiquitination ( Supplementary Figures S9A-C).
Reversely, overexpression of Parkin with ITF2 in the HeLa cells which have little or no endogenous Parkin expression, 39, 40 induced ITF2 ubiquitination (Supplementary Figure S9D).
Co-immunoprecipitation and in vitro binding assay confirmed that Parkin can bind to ITF2 ( Figures 4B, C), and it turns out to be that Parkin binds to the N-terminals of ITF2B ( Figure   4D). To corroborate which sites of ITF2 are associated with Parkin-induced ITF2 ubiquitination, we generated an ITF2 mutant where key lysine residues (Lys171-175) are removed based on the AlphaFold2 results ( Figure 4E). As expected, when the HeLa cells are overexpressed with Parkin and WT ITF2, we observed upregulation of ITF2 ubiquitination.
However, transfection with ITF2 mutant and Parkin revealed attenuated ubiquitination capacity ( Figure 4F)… In Figure 4, In Supplementary Figure S9, We added the Parkin immunoblot in Supplementary Figure S2A following the reviewer's suggestion.
In Supplementary Figure S2A Lastly, as the reviewer mentioned, Parkin has known to be activated by the Pink1 which has unique features that enable it to phosphorylate ubiquitin and the ubiquitin-like (UBL) domain of Parkin (Ser 65), and the Pink1 is reportedly activated by mitochondrial membrane potential depolarization. 3 It is unclear whether mitochondrial damage is directly associated with ITF2 loss in CAC patients. However, it is not surprising that there is a strong correlation between mitochondria dysfunction and cancer cell growth or tumorigenesis as cancers alter mitochondria functions to adapt to the surrounding environment. [4][5][6] In addition, the previous paper shows the role of mitochondria defects in IBD and colorectal cancer. 7 Taken together, mitochondria damage by the various kinds of stimuli in the colon cancer environment might modify the mitochondria membrane potential and could activate Pink1-Parkin sequentially, and then might result in ITF2 ubiquitination. In our manuscript, apart from this, we observed a huge ITF2 protein reduction in carcinoma regions of the CAC tissues and also found significant ITF2 mRNA loss in UC-associated carcinoma tissues compared to non-UC tissues in the microarray dataset ( Figure 7C). Especially, the patients who experienced severe CAC clinical symptoms (stages 3 and 4) presented much lower ITF2 levels compared to the early stages (stages 0-2) (Supplementary Figure S15B). We, therefore, might speculate that ITF2 seems to be genetically lost during the dysplasia-to-carcinoma transition as sporadic CRC did 8,9 rather than Parkin-dependent mechanisms