Targeting posttranslational modifications of RIOK1 inhibits the progression of colorectal and gastric cancers

Essential revisions:

1) Are levels of RIOK1 regulated in CRC and GC cells by the pT410/K411me switch?

Most of the experiments on SETD7/LSD1/FBXO6 mediated RIOK1 modifications and binding were done in 293T cells, typically with overexpression of epitope-tagged constructs, or in MEFs, and only in one or two cases was analysis of the endogenous SETD7/LSD1/FBXO6 proteins in RIOK1-overexpressing CRC or GC cells or tumor tissues carried out. In order to validate the physiological relevance of the phosphorylation/acetyl switch and establish that these modifications are important in the context of cancer, the authors need to use the anti-pT410 and K411me antibodies to test if pThr410 and Lys411me modifications of RIOK1 are also observed in additional CRC and GC cell lines, and assess how their levels correlate with RIOK1 protein expression levels. For instance, the authors could reproduce their findings by manipulating SETD7/LSD1/FBXO6 levels in HCT116 CRC cells, as it is known from various independent studies that they express detectable levels of endogenous RIOK1 and that its depletion has clear biological effects. Likewise, most of the interaction studies were carried out in overexpressing 293T cells, but it is not clear whether the levels of RIOK1 overexpression in transfected 293T cells are equivalent to the levels observed in CRC or GC cells, and additional data showing that these interactions occur between endogenous proteins in CRC cells need to be added to the manuscript (N.B. because so many cell types are used in various cases, the cell types tested need to be indicated by figure labels, particularly when the figures are composed of panels from multiple different cell types).

Thanks for the comments, actually we have repeated all key data in CRC and GC cell lines. Due to the large amount data already existed in this manuscript, we selectively put relevant data as suggested by the reviewers. We added Figure 9—figure supplement 1G to show the expression of pT410 and K411me in CRC cells. We also added multiple IP in Figure 5—figure supplement 1, Figure 6—figure supplement 1 and Figure 7—figure supplement 1A to show the endogenous interaction among indicated proteins in CRC cells. As suggested, we labeled the cells used in the figures.

2) Better validation of the relevance of the pT410/K411me switch to human cancer is needed.

Does RIOK1 expression or SETD7, LSD1, FBX06, and/or CK2 expression or status in CRC or GC cells correlate with any other tumor relevant features or mutations, such as RAS, AKT or tyrosine kinase mutation or activation status? To further support their conclusions, the authors should use their anti-pT410 and anti-K411me antibodies on blots of CRC and GC tissues and/or for IHC on CRC and GC tissues to correlate with RIOK1 protein levels. In this regard, since only a narrow region of the immunoblot was shown in Figure 8C, it is unclear from how specific these antibodies are for whole cell lysates, and how clean these antibodies are would dictate whether they could be used for IHC. Ideally, IHC data for these modifications should be added to the correlative data on SETD7 and CK2 expression in CRC and GC tissues shown in Figure S5.

In Figure 9, the functional relevance of K411-methylation and T410-phosphorylation was assessed using an overexpression assay in RKO rectal carcinoma cells in which RIOK1 overexpression is sufficient to enhance tumorigenicity of these RIOK1-low CRC cells. This analysis would be strengthened if the authors used mutant variants of the RIOK1-Δ RNAi resistant construct to functionally replace endogenous RIOK1 in HCT116 cells (or another RIOK1-high CRC cell line), which would indicate whether the modifications are necessary for RIOK1 function and/or regulation in CRC. Ideally, the authors would make these mutations in the endogenous RIOK1 gene using CRISPR/Cas9, but this is beyond the scope of this paper.

We didn’t check the indicated proteins expression with the status of RAS, AKT or tyrosine kinase mutation or activation status. Since we already have a lot data in this manuscript, it’s beyond the scope of current manuscript. However, it would be interesting to see these potential connections in the following manuscripts. As suggested by the reviewers, we probed pT410/K411me in CRC tissue as shown in Figure 9—figure supplement 1H. About the IHC staining, we have tried multiple times to optimize the condition, unfortunately, we got a strong background or non-specific staining in CRC or GC tissues using anti-pT410 and anti-K411me antibody. However, these two antibodies are good for WB. Further optimization of these two antibodies are needed for IHC. We agree with the comment by the reviewer “the antibodies for pT410RIOK1 and meK411RIOK1 await testing on clinical samples, which represents an interesting topic for a follow-up manuscript.” Though we could not get a clean IHC data by these two antibodies right now, we still strongly believe that there is a pT410/K411me switch among the cancer tissues based on the WB data. As suggested by the reviewers, we added relevant data of HCT116 cell line in Figure 9—figure supplement 1A, 1B, 1C, 1D, 1E and 1F.

3) The authors should discuss whether the only function of T410/K411 modification is to regulate the stability of RIOK1, or whether either of these modifications, which lie outside the kinase domain, have an independent effect on RIOK1 activity. In this regard, although elucidating the mechanistic connection between RIOK1 and AKT would be beyond the scope of the paper, it would be worth testing whether AKT inhibition leads to changes in the T410/K411 modifications or binding to any of the regulatory partners, e.g. using their antibodies to probe lysates from CRC/GC samples in the context of AKT inhibition.

As suggested by the reviewers, we detected pT410/K411me in the context of AKT inhibition, which indicates that AKT inhibition doesn’t affect the modifications of RioK1 as shown in Figure 2—figure supplement 2B.

4) The authors carried out validation of the specificity of the anti-pT410 and anti-K411me antibodies using unmodified and singly and doubly modified synthetic peptides (Figures 3F1 and S4 – in this regard the legends do not indicate exactly how these experiments were carried out, i.e. what was blotted in the different lanes/strips – immobilized peptide?).

We added how these experiments were exactly carried out and what was blotted in the different lanes/strips – immobilized peptide in the figure legends of Figure 3F1 and Figure 3—figure supplement 2.

But some concerns still remain about the specificities of the pT410 and K411me antibodies when used to detect these modifications in full length (endogenous) RIOK1 protein in cells/tissues.

We also checked these two antibodies both in cell lines and human samples, as shown in Figure 9—figure supplement 1G and 1H.

It was demonstrated that phosphorylation of T410 did not affect recognition of K411me in a synthetic peptide by the anti-K411me antibodies (Figure S4), but this is rather artificial because of the high concentration of peptide, and it remains possible that a phosphate at T410 affects the affinity of the anti-K411me antibody for K411me-modified RIOK1 protein. In fact, given what we know about the 3D structure of the CDR binding sites of antibodies that recognize linear peptide sequences, it would be surprising if a large negatively charged phosphate right next to the K411me residue does not affect binding affinity; after all pT410 prevented recognition by the SETD7 Lys methylase (in this regard, it is also unclear whether pT410 affects recognition of K411me-modified RIOK1 by the FBXO6 substrate recognition subunit of the SCF-FBXO6 E3 ligase, i.e., the experiment in Figure 6J did not test whether FBXO6 can bind to the pT410K411me doubly-modified peptide). Conversely, one might have expected that the methyl group on K411, which neutralizes the positive charge on the Lys, to affect the binding affinity of the anti-pT410 antibodies, and possibly also phosphorylation of T410 by CK2.

Thanks to the reviewer for detailed comments. About these comments, we would like to refer to one paper recently published, Fang et al., 2014. In this paper, AKT1 phosphorylates Sox2 at T118, which blocks K119me by Set7 and stabilizes Sox2. Notably, the phosphorylation site of Sox2, T118, is next to the methylation site-K119, which is consistent with our data. And our detailed data were strongly supported by experiments.

Further validation of these antibodies on the endogenous protein is really needed. For instance, any effect of T410 phosphorylation on recognition of the RIOK1 protein (rather than the RIOK1-derived peptide) could be tested by the consequences of dephosphorylating RIOK1 protein on the blot prior to antibody detection compared to a control untreated blot – there are examples in the literature of this approach (e.g. Brummer et al., 2003).

As suggested by the reviewer, we validated the antibody endogenously in HCT116 and MKN45 cell lines. Prior to development with anti-RioK1 antibodies, these cell lines were either incubated with or without λ-PPase (Figure 7—figure supplement 1B).