Co-delivery of IOX1 and doxorubicin for antibody-independent cancer chemo-immunotherapy

Anti-programmed cell death-1 (PD-1)/programmed cell death-ligand 1 (PD-L1) antibodies are currently used in the clinic to interupt the PD-1/PD-L1 immune checkpoint, which reverses T cell dysfunction/exhaustion and shows success in treating cancer. Here, we report a histone demethylase inhibitor, 5-carboxy-8-hydroxyquinoline (IOX1), which inhibits tumour histone demethylase Jumonji domain-containing 1A (JMJD1A) and thus downregulates its downstream β-catenin and subsequent PD-L1, providing an antibody-independent paradigm interrupting the PD-1/PD-L1 checkpoint. Synergistically, IOX1 inhibits cancer cells’ P-glycoproteins (P-gp) through the JMJD1A/β-catenin/P-gp pathway and greatly enhances doxorubicin (DOX)-induced immune-stimulatory immunogenic cell death. As a result, the IOX1 and DOX combination greatly promotes T cell infiltration and activity and significantly reduces tumour immunosuppressive factors. Their liposomal combination reduces the growth of various murine tumours, including subcutaneous, orthotopic, and lung metastasis tumours, and offers a long-term immunological memory function against tumour rechallenging. This work provides a small molecule-based potent cancer chemo-immunotherapy.

Potentiating Immunogenic Cell Death (ICD) is central aim of cancer immunotherapy. Jing Liu et al show that a combination of IOX1 (a demethylase and 2-OG oxygenase inhibitor) and Doxorubicin (DOC) treatment (with/without encapsulation in liposomes) resulted in increased apoptosis of cancer cells in vitro and better tumor clearance in vivo in the rather immunogenic CT26 and less immunogenic 4T1 tumor models. Overall, IOX1 improved ICD potential of DOX by reducing gene and protein expression of PD-L1 (a bona-fide checkpoint inhibitor) on cancer cells. A decreased expression of PD-L1 was accompanied by increased DC maturation, T cell infiltration and increased expression of granzyme B from tumor infiltrating T cells. Authors also report a reduction in number of tumor resident regulatory T cells. A combination of IOX1 and DOX not only cleared existing tumors, but also prevented growth of tumors upon re-introduction of CT26 cancer cells in mice. Overall, the findings are interesting and worth but the authors should take more into consideration the mechanistic insights explaining the improved antitumor effects of the combination IOX1+DOX. In the absence of more mechanistic details, some of the conclusions the authors draw seem overstatements.

Main Comments:
The study unfortunately lacks some mechanistic insights at different levels. What is the actual concentration of IOX1 once it is incorporated into the liposomes? What is the molecular mechanism of the increased cytotoxicity of IOX1+DOX? Is it simply by increasing DOX concentration in the cells by blockade of the P-gp? The author claims that this correlates with increased LC3 granularity but what is the connection with autophagy? Is this an attempt to resist to excessive cellular stress or is mechanistically involved in cytotoxicity? The conclusion that "IOX1 or its liposome formulation induces Immunogenic apoptosis due to the promoted autophagy" is not supported by the data the authors show. Related to this; the increase in LC3B area has been directly linked to activation of autophagy, but it could merely be the result of an inhibition of the lysosomal degradation of autophagosomes. Immunofluorescence images of Fig. 1 show that upon DOX+IOX1 and IDOXIL the intracellular level of CRT is increased, not per se its surface redistribution, as shown in Fig 1J. Increased levels of CRT could be a sign of the induction of ER stress and activation of the UPR. Which is a hallmark of ICD. It would be important to check whether the combo DOX+IOX1 increases ICD hallmark because of exacerbated ER stress (particularly PERK pathway). Figure 2: Was effect of IOX1 on isolated PBMCs tested? In figure 2i, although authors show that apoptosis in CT26 cells was increased upon co-incubation with PBMCs, it would be important to know if proliferation in T cells (in figure 2g) was driven by signals originating from CT26 cells or IOX1. Figure 2D: Can the author please explain why IDOXIL treatment in CT26 on western blotting shows a decrease of PD-L1 on total cell lysate but an increase on flow cytometry? To conclude that DAMPs are important for DC maturation, authors should assess the effects of the neutralization of DAMPs in their co-incubation assays, e.g. CRT by neutralizing antibody ecc. Figure 4. This reviewer finds it hard to believe that in PBS or IOXIL stained tumors authors cannot detect CRT. CRT is a housekeeping protein, so it is constitutively expressed by all cells. It has also been quite hard in the field of ICD, to distinguish the surface expression (bona fide marker of ICD) from the intracellular ER luminal localization of CRT exactly because of the overall staining of CRT in cancer cells (and all cells). Can the authors explain the lack of CRT immunoreactivity in untreated tumors? Authors show accumulation of IOX1L and IDOXIL in the cancer cells in vitro, but not in vivo, it is assumed that increased TUNEL is due to increased cancer cell death, but it could be the results of CTL activity. In vivo, authors measure some very interesting effects, including increased cell death (TUNEL) and block of tumor regrowth by the IDOXIL, correlating with increased infiltration of CTL and Grz staining and the induction of immunological memory, but actually the authors do not show whether this is an effect caused by directly forcing ICD (see comment about CRT staining) in cancer cells. For example, can the author exclude direct effects on immune cells? e.g. TAMs or T cells? Can IOXIL or IDOXIL alter directly the metabolic-epigenetic axis in T cells thus changing their activity status? Finally, in many figures it is not very clear whether N=3 means independent biological repeats? The very small SD reported, would rather suggest that they are 3 technical repeats. Also pls indicate which statistical test has been applied. One-way Anova and not T test should be applied for multiple comparisons.
Minor Figure 1f: please show a data bar for IOX1 only instead of combination with DOX regarding P-gp expression. 2e: GAPDH western blot seems overexposed. Were gene or protein expression of another checkpoint inhibitor like CTLA4 quantified? This would be important to conclude if anti-tumor activity was due to PD-L1 inhibition only. Supplementary figure 11 -images seem to be overexposed It is mentioned that small molecules could be more cost-effective in comparison to antibody-based therapy. What evidence do you have to support the potential cost-effectiveness superiorty of small molecule inhibitors over antibodies? The mentioned references mention the indeed high cost of antibodies (24) and biological efficacy of some small molecule compounds, not the economical efficacy (25). For example the small molecule inhibitor palbociclib, discussed here: doi: 10.1093/annonc/mdx201, shows 0% chance of being cost effective when being added to breast cancer therapy, with a high cost per QALY gained. Regarding the ICB induced type I diabetes, this is a relatively rare complication affecting 0,9% of patients treated with ICB (https://doi.org/10.1016/S2213-8587(19)30072-5) However useful to report pancreatic PD-L1 (maybe in regard to upcoming pancreatic cancer treatment?), lung PD-L1 might be just as useful from the sole regard of systemic toxicity since pneumonitis (and others) are slightly more frequently side effects of ICB's in general (https://doi.org/10.1002/cncr.31043).

Reviewer #2 (Remarks to the Author): with expertise in IOX1
Manuscript by Liu et al describes the potential use of small molecule IOX1 in cancer immunotherapy. The authors show IOX1 to effectively and selectively downregulate tumour PD-L1 expression in vitro and in vivo. IOX1 inhibited cancer multidrug resistance and enhanced DOXinduced ICD which remodelled the tumour-immunosupressive microenvironment including increased DC maturation and T-cell infiltration and activity, and reduced Treg cells. IOX1/DOX liposomes eradicated murine tumours, including subcutaneous, orthotopic and lung metastasis tumours, and offered long-term immunological memory function against subcutaneous and lung metastasis re-challenging.
Overall the manuscript is well written and the studies well designed. While the results are somewhat surprising given the promiscuous nature of IOX1 (inhibits a wide-range of 2OG oxygenases and likely other targets), solid data support their observations and important conclusions. IOX1 is a small molecule (better tumour penetration) and in the form of IOXIL appears to be a low-cost and safe alternative to anti-PD1/PD-L1 cancer immunotherapy in combination with DOX, thus presents a novel and exciting finding. However, the mechanism and application of IOX1-mediated increased intracellular DOX levels and reduced PD-L1 mRNA levels in cancer requires more characterisation; understanding the mechanism of action would be essential in exploiting the findings of this study.
-What is the mechanism of IOX1-mediated increased intracellular levels of DOX in CT26 cells? -Enhanced DOX cytotoxicity at cellular IC50 16nM IOX1 suggests a mechanism independent of KDM / 2OG oxygenase inhibition (e.g. most IC50s are high uM to observe significant changes in histone methylation levels / HIF upregulation) -Does IOX1 inhibit recombinant P-gp ATPase in vitro activity? -Does IOX1 increase intracellular DOX in mutant P-gp CT26 cells? -Why are there variations in IOX1-mediated enhanced DOX cytotoxicity between CT26, 4T1 & B16F10 cell lines? -is it associated with cell line specific P-gp expression / activity? -Why did IOX1 not enhance DOX cytotoxicity in NIH-3T3 and HUVEC non-cancer cell lines? -Is the P-gp expression levels / activity in NIH-3T3 and HUVEC similar to CT26? -Does IOX1 increase intracellular levels of other chemotherapeutic drugs?
Minor point Line 76: 'histone demethylase inhibitor used for the treatment of b-thalassemia30'; This paper reports the potential for use as treatment -IOX1 is not used as a treatment of b-thalassemia Reviewer #3 (Remarks to the Author): with expertise in liposomes and Doxorubicin (drug delivery) The manuscript by Liu and colleagues describes the use of a nanoliposomal histone demethylase inhibitor to potentiate the anticancer activity of liposomal doxorubicin in mice. The authors demonstrate that encapsulated IOX1 can increase doxorubicin accumulation and down regulate PD-L1 expression in mouse cancer cells. In vitro and in vivo results demonstrate impressive antitumor activity of the combined treatment of immunocompetent mice with liposomal doxorubicin and liposomal IOX1. Combined treatment stimulates DC maturation and anti-tumor T cell responses.
The manuscript is mostly well and clearly written and experiments are documented with a large amount of supplemental data. Three major points should be addressed and some minor issues need to be fixed before publication.
1. No data is shown for human cancer cells. Since gene regulation may differ between human and mouse cells, it is critical that the authors show at a minimum that IOX1 can increase dox accumulation and decrease PD-L1 expression in several human cell lines. This paper is only interesting if the results can be translated to humans.
2. The authors tend to overstate the mechanism by which antitumor activity is enhanced. For example, no direct evidence is provided showing that IOX1 affects pgp function. Investigation of cells that overexpress or have pgp KO would be more convincing. Along the same lines, although IOX1 can decrease PD-L1 expression on cancer cells, it probably alters the expression of many genes. The results show a correlation, but do not prove that the results are due to PD-L1 downregulation. Cells with PD-L1 KO would be useful to prove this. More care in claiming mechanism can solve this problem.
3. Most of the in vitro results use 5 uM DOX and 25 uM IOX1. It is unclear if these doses can be achieved in vivo. The authors do provide tumor concentrations but this appears to be free plus encapsulated drugs. Some discussion or additional data addressing the dose effects of the observations might be useful.
Minor points 1. Probably should add "in mice" to the title to make it clear the results are limited to mice.
2. The first sentences of results is misleading. "IOX1 is used for the treatment of b-thalassemia" makes it sound like this is a clinical drug. I believe the results for IOX1 are all in experimental models so far.
3. What anti-PD-L1 antibody was used? Was it specific for mouse PD-L1? 4. The authors do NOT use DOXIL in their study. LIBOd is a generic form of PEGylated liposomal doxorubicin which has a different lipid composition and mean size in comparison to DOXIL. All mentions of DOXIL should be replaced with PLD or LIBOd®.
5. The DSL data show that IOX1L ranges from ~ 50 nm to 200 nm. How can it be 102.3 +/-0.7 nm?
6. There is a lot of data showing particle size stability of the liposomes. This is relatively noninformative. The authors should show drug stability inside the liposomes.
7. The panels are placed in non-intuitive locations in some of the figures. For example, panel f in Figure 1. A more natural flow of panels from right to left and top to bottom is suggested.
8. The effect of IOX1 on rhodamine 123 accumulation in cells appears to be much more dramatic than the effect on DOX. What does that say about the proposed mechanism of action for DOX? 9. The numbers in Figure 1G  10. There are too many significant digits for the IC50 values in 1g. Also, no SD values.
11. In the in vitro studies, how do the authors know that some IOX1L is not carried over to DC, and that this is what is causing DC maturation markers to change? 12. How do the authors know that IOX1L are not directly acting on TIL in vivo?
13. The cancer cell lines should be checked for mycoplasma contamination.

Point by point response:
We are grateful to receive valuable comments on our manuscript. We appreciated all three experienced reviewers for their positive reception of our work and their constructive suggestions. Based on the reviewers' suggestions, we performed additional experiments and revised our manuscript accordingly. The major changes are highlighted in yellow. We hope the revision is clear enough to address the major point raised by the reviewers. As detailed below, we addressed each specific point raised by the reviewers.

Reviewer # 1
Overall comments: ……Overall, the findings are interesting and worth but the authors should take more into consideration the mechanistic insights explaining the improved antitumour effects of the combination IOX1+DOX. In the absence of more mechanistic details, some of the conclusions the authors draw seem overstatements. Response: Thank you for your recognition of our work and valuable suggestions. We elucidated the mechanisms, as required. Please find our itemized responses below.

Q1) What is the actual concentration of IOX1 once it is incorporated into the liposomes?
Response: The IOX1 concentrations in IOXL and IPLD were 1 mg ml -1 and 0.75 mg ml -1 respectively. The information is added in the supplementary materials.

Q2) What is the molecular mechanism of the increased cytotoxicity of IOX1+DOX? Is it simply by increasing DOX concentration in the cells by blockade of the P-gp? The author claims that this correlates with increased LC3 granularity but what is the connection with autophagy? Is this an attempt to resist to excessive cellular stress or is mechanistically involved in cytotoxicity?
The conclusion that "IOX1 or its liposome formulation induces Immunogenic apoptosis due to the promoted autophagy" is not supported by the data the authors show. Related to this; the increase in LC3B area has been directly linked to activation of autophagy, but it could merely be the result of an inhibition of the lysosomal degradation of autophagosomes. Response: We thank the reviewer's important instruction. ➢ For P-gp: In this revision, we conducted experiments and elucidated the mechanism: IOX1 downregulated P-gp expression in a dose-dependent manner through JMJD1A/β-catenin/P-gp signaling pathway (Figs. 1c,3, and Supplementary Fig. 7). IOX1 also directly inhibited the Pgp function as demonstrated in P-gp ATPase inhibition study (Supplementary Fig. 8). Thus, the increased cytotoxicity is due to the elevated intracellular DOX accumulation resulting from the P-gp downregulation and function inhibition.
➢ For autophagy: DOX can induce autophagy in cancer cells 1 , resulting in apparent LC3 signals ( Supplementary Fig. 19). On the contrary, IOX1 treatment did not change the LC3 levels in CT26 cells, suggesting IOX1 alone neither increase the autophagic flux nor inhibit the lysosomal degradation of autophagosomes. However, 5IOX1+DOX treatment increased the LC3 granularity and autophagy-associated ATP secretion 2 (Fig. 1i) in CT26 cells, indicating the increased LC3 area was caused by the IOX1-mediated enhanced intracellular DOX levels, i.e., excessive cellular stress. Fig. 1 show that upon DOX+IOX1 and IPLD the intracellular level of CRT is increased, not per se its surface redistribution, as shown in Fig 1J.

Increased levels of CRT could be a sign of the induction of ER stress and activation of the UPR, which is a hallmark of ICD. It would be important to check whether the combo DOX+IOX1 increases ICD hallmark because of exacerbated ER stress (particularly PERK pathway).
Response: We are grateful for your valuable suggestion. ➢ We isolated membrane protein and found the membrane CRT protein level was much higher after treating with 5IOX1+DOX, confirming the CRT membrane translocation.
➢ As suggested, we analyzed activation of PERK/eIF2α pathway and found that 5IOX1+DOX or their liposomes did induce much stronger ER stress than DOX or its liposome alone. The data are added in Fig. 1j and Supplementary Fig. 22.

Q4) Figure 2: Was effect of IOX1 on isolated PBMCs tested? In figure 2i, although authors show that apoptosis in CT26 cells was increased upon co-incubation with PBMCs, it would be important to know if proliferation in T cells (in figure 2g) was driven by signals originating from CT26 cells or IOX1.
Response: Thank you for your suggestion.
➢ As suggested, we tested the effect of IOX1 on isolated PBMCs (added as Supplementary Fig.   40). IOX1 had no direct effect on the proliferation index (PI) of PBMCs, indicating that the enhanced proliferation of PBMCs was driven by CT26 cells (with reduced PD-L1 expression) rather than IOX1 itself. Figure 2D: Can the author please explain why IPLD treatment in CT26 on western blotting shows a decrease of PD-L1 on total cell lysate but an increase on flow cytometry? Response: Thank you for your comment. ➢ The difference of PD-L1 amounts in western blotting and flow cytometry analysis is caused by the two methods detecting PD-L1 in different locations of cells: western blotting determines the total PD-L1 of the whole cells but flow cytometry only determines the PD-L1 on the cell membrane. To confirm this, we treated cells with Triton X-100 to make the cell membrane permeable enabling anti-PD-L1 antibody to label the intracellular PD-L1 as well. The flow cytometry analysis of such cells now is consistent with the results of western blotting. Please see the results in Fig. 2e and Supplementary Fig. 35.

Q6)
To conclude that DAMPs are important for DC maturation, authors should assess the effects of the neutralization of DAMPs in their co-incubation assays, e.g. CRT by neutralizing antibody ecc. Response: Thank you for your advice. ➢ We used an alternative to answer the reviewer's question according to the reference method 3 .
CT26 cells were deprived of CRT by its shRNA and then treated with IOX1+DOX. These cells hardly caused DC maturation (Supplementary Figs. 28,29), proving that CRT, one of the DAMPs, is necessary to DC maturation. Figure 4. This reviewer finds it hard to believe that in PBS or IOXL stained tumours authors cannot detect CRT. CRT is a housekeeping protein, so it is constitutively expressed by all cells. It has also been quite hard in the field of ICD, to distinguish the surface expression (bona fide marker of ICD) from the intracellular ER luminal localization of CRT exactly because of the overall staining of CRT in cancer cells (and all cells). Can the authors explain the lack of CRT immunoreactivity in untreated tumours? Response: Thanks for your question. ➢ The CRT expression in tumours of PBS or IOXL-treated mice could be detected but was very weak compared with PLD and IPLD groups. Therefore, in the images given in Fig. 5a, the signals in the PBS and IOX1 groups are hardly seen. As you can see from the following images (Fig. R), once the immunofluorescence in the two groups is seen clearly, those in the PLD and IPLD groups would be overexposed.

Q8)
Authors show accumulation of IOXL and IPLD in the cancer cells in vitro, but not in vivo, it is assumed that increased TUNEL is due to increased cancer cell death, but it could be the results of CTL activity. Response: Thank you for your advice. ➢ CTLs did exert strong effect on the excellent antitumour efficiency. The increased TUNEL was attributed to the CTLs activity in the tumours as well.

Q9) In vivo, authors measure some very interesting effects, including increased cell death (TUNEL) and block of tumour regrowth by the IPLD, correlating with increased infiltration of CTL and Grz staining and the induction of immunological memory, but actually the authors do not show whether this is an effect caused by directly forcing ICD (see comment about CRT staining) in cancer cells. For example, can the author exclude direct effects on immune cells? e.g. TAMs or T cells? Can IOXL or IPLD alter directly the metabolic-epigenetic axis in T cells thus changing their activity status?
Response: Thank you for your suggestion. ➢ Following your advice, we measured the percentage of M2 macrophages ( Supplementary Fig.   64) in tumours after different treatments and found that both IOXL and IPLD had no influence on the contents of M2 macrophages in the tumours.
➢ We demonstrated that IOX1 inhibited JMJD1A and thus downregulated its downstream βcatenin and subsequent PD-L1 and P-gp expression. The JMJD1A in T lymphocytes was much lower than that in CT26 cells (Supplementary Fig. 48). Therefore, IOX1 can not affect the metabolic-epigenetic axis of T cells. The experiments also confirmed that IOX1 had no direct effect on the proliferation of T cells (Supplementary Fig. 40).

Q10) Finally, in many figures it is not very clear whether N=3 means independent biological repeats?
The very small SD reported, would rather suggest that they are 3 technical repeats. Also pls indicate which statistical test has been applied. One-way Anova and not T test should be applied for multiple comparisons. Response: Thank you for your advice. ➢ During the studies, all the cell experiments were repeated at least 3 times independently. For animal experiments, the samples were from at least 3 different mice in every group. We