Ketohexokinase-A acts as a nuclear protein kinase that mediates fructose-induced metastasis in breast cancer

Harmful effects of high fructose intake on health have been widely reported. Although fructose is known to promote cancer, little is known about the underlying mechanisms. Here, we found that fructose triggers breast cancer metastasis through the ketohexokinase-A signaling pathway. Molecular experiments showed that ketohexokinase-A, rather than ketohexokinase-C, is necessary and sufficient for fructose-induced cell invasion. Ketohexokinase-A-overexpressing breast cancer was found to be highly metastatic in fructose-fed mice. Mechanistically, cytoplasmic ketohexokinase-A enters into the nucleus during fructose stimulation, which is mediated by LRRC59 and KPNB1. In the nucleus, ketohexokinase-A phosphorylates YWHAH at Ser25 and the YWHAH recruits SLUG to the CDH1 promoter, which triggers cell migration. This study provides the effect of nutrition on breast cancer metastasis. High intake of fructose should be restricted in cancer patients to reduce the risk of metastasis. From a therapeutic perspective, the ketohexokinase-A signaling pathway could be a potential target to prevent cancer metastasis.

Reviewer #2, Expertise: Breast cancer, EMT and metastasis (Remarks to the Author): In this manuscript, the authors described a new mechanism by which KHK-A mediates the fructose-induced epithelial-mesenchymal transition and invasion of breast cancer cells. Mechanistically, KHK-A interacts with LRRC 59 and KPNB1 in fructose dependent manner, and thereby is translocated into nucleus where it phosphorylates YWHAH. Phosphorylated YWHAH recruits Slug to the promoter region and represses the transcriptional activity of C DH1 gene. The experiments are well-designed with rigorous controls, and the data well support their hypothesis. However, the clinical evidence is lacking and the physiological significance of KHK-A signaling is not fully addressed, which are needed to better justify the translational potential of their findings. Major comments: 1) The relationship between fructose consumption and metastatic progression of breast cancer has not been observed or reported, though such connection has been studied on mouse models. The lacking of clinical evidence of the fructose/KHK-A/YWHAH axis in promoting breast cancer metastasis is the major weakness of this manuscript. To justify the clinical importance of their findings, it will be better to investigate the relationship among KHK-A expression, YWHAH phosphorylation and metastasis progression in patient samples.
2) The major findings of fructose/KHK-A induced EMT, such as down-regulation of both protein and mRNA level of C DH1, and C hIP assay of SLUG binding on C DH1 promoter, were test only on MDA-MB-231 cells. MDA-MB-231 cells are well known to be an invasive breast cancer cell line with extremely mesenchymal phenotype. It is not common to further induce EMT in such cells. In addition, EMT is known to lead to the loss of epithelial markers accompanied with gain of mesenchymal markers. However, as shown in Figure 1f, neither fructose treatment alone or with over-expressed KHK-A could increase the expression levels of mesenchymal markers in MDA-MB-231 cells. Therefore, another epithelial cell line, such as MC F7, is required to validate the effects of fructose/KHK-A on EMT.
3) The current design of in vivo study is less clinically relevant. The metastasis promoting effect of fructose drink was only observed in the group with ectopic expression of KHK-A but not in the control group with the physiological level of KHK-A. It would be better to include another design showing that loss of function of KHK-A could diminish the metastasis in vivo. Minor comments: 1) Figure 1b, the absolute quantitation qRT-PC R of KHK isoforms should be used to compare KHK-A and KHK-C expression levels in cells. The relative quantitation qRT-PC R may be misleading due to the difference in the amplification efficiency of KHK-A and KHK-C primers. 2) MDA-MB-231 cells are typically spindle shaped morphology in monolayer culture (https://www.atcc.org/products/all/HTB-26.aspx#characteristics), and have high expression level of vimentin but very weak or even non-detectable protein levels of both E-C adherin and N-C adherin as determined by western blotting (PMID: 10545506). The current immunofluorescence and immunoblotting data are not consistent among different panels or with the above mentioned impression on MDA-MB-231 cells. A cell authentication report may be required to prove the identity of MDA-MB-231 used in this study.
3) Several references are not properly cited. For example, neither literature 5 nor 6 discussed the association of fructose consumption and risk of multiple types of cancer except the endometrial cancer. Literature 18 does not report that breast cancer cells can metastasize in response to fructose. 4) Both YWHAH and KHK-A proteins are also present in the cytosolic compartment, as suggested by Figure 3D. Thus, the figure 8 is not accurate and kind of misleading. The mechanism underlying the observation that KHK-A does not interact with cytosolic YWHAH upon fructose treatment may illuminate new strategies to block this signaling axis.
Reviewer #3, Expertise: molecular modelling and dynamics (Remarks to the Author): The paper investigates the molecular mechanisms at the basis of fructose intake and breast cancer metastasis, supporting clinical evidence for fructose-induced cancer aggravation. The authors present evidences that KHK-A signaling pathway is involved in fructose-induced cell invasion. In particular, they show molecular experiments supporting the evidence that ketohexokinase-A, rather than ketohexokinase-C , is necessary and sufficient for fructose-induced cell invasion.
While I find the paper interesting and well organised, with a nice combination of computational and experimental results, I have some major reservations that, I think, are in the way for a clear conclusion on the reported data.
The evidence that the main target is KHK-A and not other isoforms is based on observed overexpression of KHK-A in a number of relevant cell lines, but should be better supported. A recent paper (not cited by the authors) C lin Invest. 2018;128(6):2226-2238. https://doi.org/10.1172/JC I94427 present convincing evidence that protection from High Fructose Intolerance (HFI) is resulting from inhibition of the KHK-C isoform, and not the A. So the mechanistic interplay of these two Kinases seems to have a prominent role with respect to fructose intake. Once silenced KHK-A what happens to the expression on KHK-C in vivo? We do not know if these isoforms would replace the KHK-A activity in such condition and hterefore we cannot use the argument of specifically inhibiting KHK-A. Moreover how specific is the named inhibitor is not clarified and not texted here with respect to the frcutose signalling pathways. The authors should show the results also with lest specific inhibitors, and if available, inhibitors targeting KHK-C vs KHK-A.
I have some major reservations also on the computational data here reported. The docking and pulling simulations are very poorly described. The intent of the pulling simulation was to describe the Potential of mean force (PMF)associated to the process of S25-OH approaching ATP-gammaPO3 and therefore support the hypothesis that YWHAH is phosphorylating KHK-A (Fig  6 panel a). First of all the description of the residue contacts playing as role in this process is difficult to follow and the comparison with PRPS11 is not clearly discussed and documented. Moreover the PMF is reported as a single profile for each experiment and there is no evidence of the reproducibility or of the error associated to these. Multiple replica of the pulling process are usually needed to validate the observations. Additionally, resampling with the WHAM method, for example, would at least give some confidence to the observed overlap or differences in the profiles.  Supplementary Fig 3f and 3g. Lines 20-24 on p5; Lines 1-6 on p28; Lines 12-18 on p29.

Revision:
2. Fig.1 and Supplementary Fig. 1a. These experiments are not rationally designed. The conclusion that fructose but not glucose promotes cell invasion is misleading. It is well known that glucose is critical for tumor cell migration, invasion, and proliferation. The authors should add an equal amount of fructose and glucose into glucose-and fructose-free medium when doing these experiments.
Answer: As suggested, we added an equal amount (finally 5 mM) of fructose or glucose to glucose/fructose-free medium and performed the invasion assay. We found that breast cancer cells get more invasive under fructose stimulation but not under glucose stimulation.
Given previous reports, hyperglycemia (>25 mM glucose) seems to facilitate cancer cell migration. As this work is not related with diabetes, we here have no reason for checking the hyperglycemia effect on breast cancer metastasis. Answer: To examine the fructose flux, we measured fructose-1-phosphate and ROS levels.
The fructose-induced enhancements of fructose-1-phosphate and ROS were observed in breast cancer cells expressing KHK-C, and those were inhibited by Pypy. However, the KHK-A-expressing cells did not produce fructose-1-phosphate and ROS in response to fructose regardless of Pypy treatment. Thus, we concluded that the inhibitory effect of Pypy against invasion of KHK-A-expressing cells is not attributed to the inhibition of fructose metabolism.

Answer:
We tried to test possibility of the involvement of pre-mentioned proteins with fructose-induced breast cancer metastasis ( Supplementary Fig 4a and b). The inconsistent results with previous reports may be due to differences in cell lines and experimental conditions. Revision: Lines 3-5 on p6

KHK-A depletion with reconstituted expression of KHK-C.
Answer: We performed in vivo studies with sh-Khk-a deficient cancer cells with/ or without expressing KHK-C.

Does fructose treatment affect the binding of KHK-A to PRPS1 and PRPS1 phosphorylation by KHK-A since some of KHK-A translocates to the nucleus?
Answer: We checked these points in three different cell lines, as shown in the bottom. The phosphorylation of PRPS1 appears to be slightly reduced in breast cancer cell lines treated with fructose, but not in Hep3B hepatoma cells. We also checked the protein interaction.
Fructose does not affect the interaction between PRPS1 and KHK in all cell lines tested.
Therefore, we thought that fructose may functionally inhibit the KHK-A phosphoreaction of PRPS1 cell context-dependently. Even though we got these data, we do not want to include them in the present study focusing on YWHAH because they are not essential in this work.
Hep3B, MCF-7 and MDA-MB-231 cells were incubated with 5 mM fructose, and cell lysates were subjected to immunoprecipitation with PRPS1 antibody (or IgG), and immunoblotted with anti-KHK, or anti phospho-S/T antibody.

If leucine 83 of KHK-A is involved in binding to both PRPS1 and KPNB1/LRRC59, how does fructose regulate the binding of to KHK-A to PRPS1 versus to KPNB1/LRRC59?
Answer: Given the above results for the revision point #9, fructose does not affect the interaction between KHK-A and PRPS1. Yet, we do not know how fructose reinforces the interaction between KHK-A and KPNB1/LRRC59 in the present time. To understand the differences in fructose effect on the KHK-A binding to PRPS1 and KPNB1/LRRC59, we need to carefully review the 3D molecular structures. This remains to be investigated in the future. Fig. 5. KHK-C should be included as a control in the protein kinase assay.

11.
Answer: Recombinant His-KHK-C and GST-YWHAH peptides were reacted in the kinase reaction buffer with or without fructose. As expected, YWHAH was not phosphorylated by KHK-C.

The IC50 and Ki of fructose on inhibition of KHK-A-mediated phosphorylation of YWHAH should be measured. What is the nuclear concentration of fructose when the cells treated with fructose? Will KHK-A be inhibited by nuclear fructose under this concentration?
Answer: As suggested, we examined the enzyme kinetics for the fructose effect on the KHK-A phosphorylation of YWHAH. The Km value of KHK-A for YWHAH was 246 nM. The IC 50 and Ki values of fructose against the KHK-A phosphorylation of YWHAH was 0.91 mM and 0.72 mM, respectively. We also quantified the nuclear concentration of fructose. When breast cancer cells were incubated with 5 mM fructose, fructose was substantially detected in the cytosolic fraction, but not in the nuclear fraction, which indicates that fructose hardly enters into the nucleus. These results suggest that the fructose inhibition of YWHAH phosphorylation does not occur in the nucleus. Supplementary Fig 9e, 9f, 9g, and 9h. Lines 17-25 on p9, and Line 1 on p10;

14-3-3 (YWHAH) is also a cytosolic protein. Why is KHK-A unable to phosphorylate YWHAH in cytosol?
Answer: This is a very reasonable question. In many cases, protein modifying enzymes require some scaffold partners for its specific reaction with their substrate proteins. Now we do not precisely know whether some subunits are essential for the KHK-A reaction. If so, the KHK-A reaction with YWHAH may be activated in the presence of some scaffold proteins exclusively existing in the nucleus. This is an open question to be solved.

Fig. 5G. Expression of Flag-tagged YWHAH proteins only limitedly restored the expression of YWHAH. Why did the cells with Flag-tagged WT YWHAH expression have much more cell invasion than their parental cells?
Answer: We understand that the western blot in Fig. 5g raised such a concern. This problem arose from the different sensitivity of antibodies against the same samples. Thus, we replaced the YWHAH blots in fig 5g and supplementary Fig 9k with others. In these results, we detected the endogenous YWHAH and ectopically expressed Flag-YWHAH in the same blots. The blots clearly showed that Flag-tagged YWHAH is much more expressed than endogenous YWHAH. Fig 5g and Supplementary Fig 9k. 15. Fig. 6. Does D220 in PRPS1 abolish the binding of PRPS1 to KHK-A and PRPS1 phosphorylation by KHK-A?

Revision:
Answer: A molecular dynamics analysis showed that a region (aa.100-114) of KHK-A closely contacts with D220 in PRPS1 (bottom in Fig. 7d). Theoretically, the mutation of D220 is expected to inhibit the interaction between PRPS1 and KHK-A. However, we think we do not have to test this possibility in the present study. Actually, that is definitely out of our interest. This possibility could be tested by the original research team to study PRPS1.

Fig. 7. Does expression of YWHAH S25A blocked tumor metastasis in mouse?
Answer: For in vivo evaluation of breast cancer metastasis, murine MTV-TM-011-Luc-YWHAH_WT and YWHAH_S25A cells were injected orthotopically into the mammary fat pads of mice. Our results show metastatic tumor were detected in the lung, liver, spleen, kidney, brain and GI tract in YWHAH_WT expressing mice. However, in YWHAH_S25A bearing mice, metastatic tumors were hardly detected. Histological examination also showed chest metastasis rarely developed in YWHAH_S25A bearing mice, compared to YWHAH_WT bearing mice. Collectively, it is concluded that the KHK-A-phosphorylated residue, S25 of YWHAH is essential for the fructose-induced metastasis in breast cancer. Fig. 6. Lines 15-25 on p10, and Lines 1-3 on p11; Lines 1-10 on p20, Lines 15-20 on p20; Lines 5-18 on p36.

A recent study (Xu et al., Science Advances, 2019) showed that KHK-A phosphorylates p62 promote tumor cell survival. Does adding fructose in medium alter p62 phosphorylation by KHK-A?
Answer: I am sorry that I could not understand such a question. We have not performed any experiments related with p62. In fact, this point is absolutely out of our interest. We have a certain hypothesis and we test it, which is the science. We cannot test everything that have been done by other research groups. I think this possibility would be better tested by the original research team that are interested in p62.

Reviewer #2
In this manuscript, the authors described a new mechanism by which KHK-A mediates the fructose-induced epithelial-mesenchymal transition and invasion of breast cancer cells.
Mechanistically, KHK-A interacts with LRRC59 and KPNB1 in fructose dependent manner, and thereby is translocated into nucleus where it phosphorylates YWHAH. Phosphorylated YWHAH recruits Slug to the promoter region and represses the transcriptional activity of CDH1 gene. The experiments are well-designed with rigorous controls, and the data well support their hypothesis. However, the clinical evidence is lacking and the physiological significance of KHK-A signaling is not fully addressed, which are needed to better justify the translational potential of their findings.
Major comments: 1) The relationship between fructose consumption and metastatic progression of breast cancer has not been observed or reported, though such connection has been studied on mouse models. The lacking of clinical evidence of the fructose/KHK-A/YWHAH axis in promoting breast cancer metastasis is the major weakness of this manuscript. To justify the clinical importance of their findings, it will be better to investigate the relationship among KHK-A expression, YWHAH phosphorylation and metastasis progression in patient samples.

Answer:
To examine the association of KHK-A and S25-phosphorylated YWHAH, we immunologically stained human breast cancer arrays. The cancer specimens were histologically graded according to the Nottingham grading system. The nuclear KHK-A and S25-phosphorylated YWHAH levels both were significantly higher in the grade 3 group than in the grade 2 group. Even when the breast cancer specimens were divided into nonmetastasis and metastasis groups, the nuclear KHK-A and S25-phosphorylated YWHAH levels were much higher in the metastasis group. Spearman correlation analysis showed that nuclear KHK-A expression correlates with the S25-phosphorylated YWHAH expression.
3) The current design of in vivo study is less clinically relevant. The metastasis promoting effect of fructose drink was only observed in the group with ectopic expression of KHK-A but not in the control group with the physiological level of KHK-A. It would be better to include another design showing that loss of function of KHK-A could diminish the metastasis in vivo.
Answer: As suggested, we performed new experiments. First of all, I would like to mention that the fructose-induced metastasis was not only observed in KHK-A overexpressing breast tumors, but also in control ones. Given Supplementary figures 5b, 6e, and 10b, the lung metastasis by fructose intake was also detected in the control tumors. However, when endogenous KHK-A was silenced or replaced with ectopic KHK-C, the fructose intake failed to promote tumor metastasis.
Revision: Supplementary Fig. 6 and Fig 10. Lines 12-13, 15, 19-20 and 23-24 on p6. Lines 1-6 on p7, Line 21 on p10; Lines 15-20 on p20. Answer and Revision: Western blot results are usually influence by various factors including antibody sensitivity, dilution titer, handling of membrane, and so on. We also had hard time to detect E-cadherin and N-cadherin, but overcame this huddle after trying many different antibodies. In addition, the mRNA result (Fig. 8b) also support that the gene for Ecadherin is expressed, not completely suppressed. In terms of the identity of MDA-MB-231, we here attached the STR profile report provided by The Korea Cell Line Bank (Seoul, South Korea). The result shows that the cell line used in this study has the same STR profile to that presented at ATCC, as follows. The ATCC STR profile for MDA-MB-231 (ATCC HTB-26) is, Amelogenin: X; CSF1PO: 12, 13; D13S317: 13; D16S539: 12; D5S818: 12; D7S820: 8,9;TH01: 7,9,3;TPOX: 8,9;vWA: 15,18) Figure 3D. Thus, the figure 8 is not accurate and kind of misleading. The mechanism underlying the observation that KHK-A does not interact with cytosolic YWHAH upon fructose treatment may illuminate new strategies to block this signaling axis.

Revision:
We revised the graphical summary in Fig. 10, as suggested.

Reviewer #3
The paper investigates the molecular mechanisms at the basis of fructose intake and breast cancer metastasis, supporting clinical evidence for fructose-induced cancer aggravation. The authors present evidences that KHK-A signaling pathway is involved in fructose-induced cell invasion. In particular, they show molecular experiments supporting the evidence that ketohexokinase-A, rather than ketohexokinase-C, is necessary and sufficient for fructoseinduced cell invasion.
While I find the paper interesting and well organised, with a nice combination of computational and experimental results, I have some major reservations that, I think, are in the way for a clear conclusion on the reported data.
1) The evidence that the main target is KHK-A and not other isoforms is based on observed overexpression of KHK-A in a number of relevant cell lines, but should be better supported.
Answer: To clarify distinct function of KHK-A from KHK-C in breast cancer metastasis, we additionally performed experiments with breast cancer cell line overexpressing KHK-C both in vitro and in vivo: Revision: Fig 1d, 1e, Supplement Fig 3f, 3g, 6, 7b, 7c, 8b, 8e, and 9d. Lines 5-8 on p5 For the second question, the inhibitor used in this project is pyrimidinopyrimidine (also known as KHK inhibitor, sigma, 420640). According to the information sheet, this comical is a reversible and ATP-competitive inhibitor. IC50 against KHK is so low as 12 nM, which may support its specificity to KHK. The inhibitor targets the ATP binding motif, which is 100% conserved in KHK-A and KHK-C, so it inhibits both isoforms with the same efficiency. As far as I know, unfortunately, the isoform-specific inhibitors have not been available yet.

3) Major revision
I have some major reservations also on the computational data here reported.
The docking and pulling simulations are very poorly described. The intent of the pulling simulation was to describe the Potential of mean force (PMF)associated to the process of S25-OH approaching ATP-gammaPO3 and therefore support the hypothesis that YWHAH is phosphorylating KHK-A (Fig 6 panel a). First of all the description of the residue contacts playing as role in this process is difficult to follow and the comparison with PRPS11 is not clearly discussed and documented. Moreover the PMF is reported as a single profile for each experiment and there is no evidence of the reproducibility or of the error associated to these. Multiple replica of the pulling process are usually needed to validate the observations. Additionally, resampling with the WHAM method, for example, would at least give some confidence to the observed overlap or differences in the profiles.
Answer: Upon the reviewer's comments, we added more descriptions for docking and pulling simulations (in Method section for Molecular dynamics simulation) and elaborated more on the discussion for the comparison with PRPS11 and for the clarification of contact map. Also, following the reviewer's comments on the reproducibility of PMF, we performed 5 fully independent sets of umbrella simulations for each of profiles to check the reproducibility.
The uncertainty of each profile was also estimated by resampling technique using bootstrap analysis with 200 bootstraps. Fig. 7a. Lines 2-11 on p12.