Microbial Metabolites Potentiate MAIT Cell Anti-Tumor Immunity Against Solid Tumors

Reviewer #1 (Public review):

The manuscript from Zhu et al. identifies microbial riboflavin-derived MR1 ligands as potent pharmacological activators of human MAIT cells and provides evidence that MR1 ligand stimulation can enhance MAIT-mediated tumor killing across multiple solid tumor models. The study is conceptually interesting and supported by a broad combination of human primary samples, tumor cell lines, 3D models, SC transcriptomics, and xenograft experiments. Overall, the data largely support the central conclusion that MR1 ligand stimulation can strongly activate human MAIT cells and enhance anti-tumor cytotoxicity. However, the broader conclusions concerning endogenous MAIT mobilization, tumor specificity, and translational potential are not yet fully supported by the current data and should either be moderated or addressed with additional experiments.

Comments:

(1) The authors use one-way ANOVA throughout the manuscript, but this may not be appropriate for some analyses, particularly when multiple experimental factors are present and their interaction effects need to be considered. For example, Figure 3f appears to involve multiple factors, for which a two-way ANOVA may be more appropriate. Similar issues may apply to other panels.

(2) In Figure 3f, the authors show data from patients #1 and #2 and state that the experiment is representative of three experiments. What does the reported "n=4" represent in this figure?

(3) There appears to be a discrepancy between Figure 3f and Supplementary Figure 3b. The two panels appear to use the same treatment conditions and the same label, and both appear to use patient #1 samples, yet the reported values are different. Please clarify the experimental design and explain the reason for this discrepancy.

In addition, the gating strategy used to define live tumor cells should be clearly described in the figure legend and/or Methods. The authors define "live tumor cells" as MR1/5-OP-RU tetramer-CD45- cells. However, in primary liver tumor samples, the CD45-/tetramer- population may contain other non-hematopoietic cells, such as fibroblasts, and therefore may not exclusively represent tumor cells. The authors should clarify whether additional tumor-specific markers or other criteria were used. The gating strategies for the relevant flow cytometry experiments should be provided in the Supplementary figures.

(4) I have some concerns regarding the claims of "selective activation of anti-tumor inflammatory pathways rather than generalized cytokine release" and "avoiding induction of tumor-supportive mediators." The authors show that MAIT cells stimulated with 5-OP-RU can substantially reduce tumor cell viability. Therefore, the cellular composition of the co-culture is likely to change considerably during the assay, which may affect the absolute levels of cytokines and other soluble mediators detected. For example, reduced tumor cell numbers could lead to lower production of tumor-derived factors such as VEGF, potentially confounding the interpretation that these mediators are not induced by MAIT activation. The authors should consider whether cytokine measurements have been normalized to viable cell numbers or otherwise account for differences in tumor cell abundance.

(5) The in vivo tumor models may show substantial variability between independent experiments. Rather than presenting a single representative experiment, the authors should consider showing pooled data from all independent experiments, with the total number of mice clearly indicated.

(6) Why did the authors use an MR1-overexpressing tumor cell line for the in vivo studies rather than the parental cells with endogenous MR1 expression, together with MR1-KO cells as a negative control? The authors demonstrate that MR1 is detectable across multiple tumor cell lines and that endogenous MR1 expression is sufficient to support MAIT-mediated killing in vitro. Moreover, MR1 overexpression substantially enhances tumor cell susceptibility to MAIT-mediated killing. Therefore, it is unclear whether the strong therapeutic efficacy observed in vivo reflects physiologically relevant MR1 expression or is driven by artificially elevated MR1 expression. An in vivo comparison using parental and MR1-KO tumor cells would substantially strengthen the translational relevance and establish whether the therapeutic effect can be achieved at endogenous levels of MR1.

(7) How is tumor specificity of MAIT achieved ? The authors propose that MAIT-cell activation by MR1 ligands provides an antigen-independent approach for tumor targeting. However, MR1 is broadly expressed and is not tumor specific. While the relative sparing of T and B cells in Figure 7B provides some evidence of cell-type selectivity, this does not establish tumor versus normal tissue specificity. It remains unclear whether activated MAIT cells can discriminate tumor cells from other normal MR1-expressing cells and tissues. This raises an important question regarding the potential systemic toxicity of MAIT cells activated by systemic administration of 5-OP-RU. In particular, could other MR1-expressing cells be targeted when a large number of MAIT cells are simultaneously activated? The authors should consider assessing systemic toxicity in vivo, for example by examining serum ALT/AST levels and tissue pathology, and/or by evaluating the effects of MAIT + 5-OP-RU in tumor-free animals. At least, the potential specificity and safety limitations of systemic MR1 agonism should be discussed.

Reviewer #2 (Public review):

The manuscript by Zhu et al. describes MAIT cell activation by riboflavin metabolites presented by MR1. The authors provide solid evidence for this activation and anti-cancer functional consequence using an array of selected cell lines, primary ex vivo and engineered xenograft models. Broadly, the results are thorough and well controlled, and provide a highly informative insight into the metabolite-MAIT-cancer cell interactions. However, the majority of this work is undertaken using models that preferentially express key targets, and whilst still useful, the (current) broader implications of this research are overstated. Additionally, the suggested MAIT modulation of the tumor microenvironment requires clarification.

Major Comments:

(1) In Figures 2b-d, the authors suggest microbial metabolite stimulation of PBMC cultures increased MAIT cell frequency up to 60%. Whilst their flow data is compelling, the frequency of one population can be influenced by changes in other populations. A form of absolute or relative-to-total count should be used.

(2) The statements regarding cytokine induction in Figure 4e are too strong; many of those inflammatory cytokines are not automatically and consistently tumour-suppressive. The line 299 '...were not induced' may just reflect death of tumor cells. It would be useful to include tumour cell-only controls in Figure 4.

(3) Figure 7 is interesting, but the authors' conclusion that MAIT+5-OP-RU controls the tumor microenvironment is not robustly supported by their evidence.

a) It is not clear how CD14+ cells established a sustained suppressive environment.

b) It is not clear how the peritoneal addition of microbial metabolites 'significantly enhanced MAIT-mediated tumor control'. The authors show that the addition of 5-OP-RU reduced the number of GFP-expressing tumour cells present in peritoneal lavage fluid. There is limited evidence to suggest this occurs through MAIT cells or MR1 in this figure.

c) It is difficult to draw conclusions from peritoneal lavage flow when some experimental groups received cells IP, but then all groups were equally assessed for key populations, and all data are presented as frequencies. The authors should use absolute counts (or similar) to appropriately show changes in cell populations to account for varying total/live/cd45+ cell compartments.

d) It would be necessary at a minimum to include 5-OP-RU-only controls, and ideally include MR1 blocking or the cancer line with MR1 removed. Alongside this, the authors should substantially reduce the strength of their statements on microbial metabolite-MAIT suppression of the tumor microenvironment.

Author response:

Reviewer #1 (Public review):

The manuscript from Zhu et al. identifies microbial riboflavin-derived MR1 ligands as potent pharmacological activators of human MAIT cells and provides evidence that MR1 ligand stimulation can enhance MAIT-mediated tumor killing across multiple solid tumor models. The study is conceptually interesting and supported by a broad combination of human primary samples, tumor cell lines, 3D models, SC transcriptomics, and xenograft experiments. Overall, the data largely support the central conclusion that MR1 ligand stimulation can strongly activate human MAIT cells and enhance anti-tumor cytotoxicity. However, the broader conclusions concerning endogenous MAIT mobilization, tumor specificity, and translational potential are not yet fully supported by the current data and should either be moderated or addressed with additional experiments.

We thank the reviewer for the positive feedback. We will address all comments and suggestions point by point.

Comments:

(1) The authors use one-way ANOVA throughout the manuscript, but this may not be appropriate for some analyses, particularly when multiple experimental factors are present and their interaction effects need to be considered. For example, Figure 3f appears to involve multiple factors, for which a two-way ANOVA may be more appropriate. Similar issues may apply to other panels.

We thank the reviewer for this valuable comment. We will carefully review the statistical analyses and revise the tests as appropriate, including the use of two-way ANOVA where multiple experimental factors are present.

(2) In Figure 3f, the authors show data from patients #1 and #2 and state that the experiment is representative of three experiments. What does the reported "n=4" represent in this figure?

We thank the reviewer for this valuable comment. We will revise the figure legend to clearly define what the reported n = 4 represents.

(3) There appears to be a discrepancy between Figure 3f and Supplementary Figure 3b. The two panels appear to use the same treatment conditions and the same label, and both appear to use patient #1 samples, yet the reported values are different. Please clarify the experimental design and explain the reason for this discrepancy.

In addition, the gating strategy used to define live tumor cells should be clearly described in the figure legend and/or Methods. The authors define "live tumor cells" as MR1/5-OP-RU tetramer-CD45- cells. However, in primary liver tumor samples, the CD45-/tetramer- population may contain other non-hematopoietic cells, such as fibroblasts, and therefore may not exclusively represent tumor cells. The authors should clarify whether additional tumor-specific markers or other criteria were used. The gating strategies for the relevant flow cytometry experiments should be provided in the Supplementary figures.

We thank the reviewer for this valuable comment. Figure 3f (patient #2) and Supplementary Figure 3b (patient #1) were generated using samples from different patients. We will clarify this in the revised manuscript and provide the relevant gating strategies in the Supplementary Information.

(4) I have some concerns regarding the claims of "selective activation of anti-tumor inflammatory pathways rather than generalized cytokine release" and "avoiding induction of tumor-supportive mediators." The authors show that MAIT cells stimulated with 5-OP-RU can substantially reduce tumor cell viability. Therefore, the cellular composition of the co-culture is likely to change considerably during the assay, which may affect the absolute levels of cytokines and other soluble mediators detected. For example, reduced tumor cell numbers could lead to lower production of tumor-derived factors such as VEGF, potentially confounding the interpretation that these mediators are not induced by MAIT activation. The authors should consider whether cytokine measurements have been normalized to viable cell numbers or otherwise account for differences in tumor cell abundance.

We thank the reviewer for this important comment. We agree that differences in tumor cell abundance may affect cytokine measurements. We will moderate our claims accordingly and acknowledge this limitation in the revised manuscript.

(5) The in vivo tumor models may show substantial variability between independent experiments. Rather than presenting a single representative experiment, the authors should consider showing pooled data from all independent experiments, with the total number of mice clearly indicated.

We thank the reviewer for this valuable comment. We will provide pooled data from all independent in vivo experiments and clearly indicate the total number of mice.

(6) Why did the authors use an MR1-overexpressing tumor cell line for the in vivo studies rather than the parental cells with endogenous MR1 expression, together with MR1-KO cells as a negative control? The authors demonstrate that MR1 is detectable across multiple tumor cell lines and that endogenous MR1 expression is sufficient to support MAIT-mediated killing in vitro. Moreover, MR1 overexpression substantially enhances tumor cell susceptibility to MAIT-mediated killing. Therefore, it is unclear whether the strong therapeutic efficacy observed in vivo reflects physiologically relevant MR1 expression or is driven by artificially elevated MR1 expression. An in vivo comparison using parental and MR1-KO tumor cells would substantially strengthen the translational relevance and establish whether the therapeutic effect can be achieved at endogenous levels of MR1.

We thank the reviewer for this important comment. We agree that comparison with endogenous MR1 expression would strengthen the translational relevance of our findings. We will include new in vivo experiment comparing parental tumor cells.

(7) How is tumor specificity of MAIT achieved? The authors propose that MAIT-cell activation by MR1 ligands provides an antigen-independent approach for tumor targeting. However, MR1 is broadly expressed and is not tumor specific. While the relative sparing of T and B cells in Figure 7B provides some evidence of cell-type selectivity, this does not establish tumor versus normal tissue specificity. It remains unclear whether activated MAIT cells can discriminate tumor cells from other normal MR1-expressing cells and tissues. This raises an important question regarding the potential systemic toxicity of MAIT cells activated by systemic administration of 5-OP-RU. In particular, could other MR1-expressing cells be targeted when a large number of MAIT cells are simultaneously activated? The authors should consider assessing systemic toxicity in vivo, for example by examining serum ALT/AST levels and tissue pathology, and/or by evaluating the effects of MAIT + 5-OP-RU in tumor-free animals. At least, the potential specificity and safety limitations of systemic MR1 agonism should be discussed.

We thank the reviewer for this important comment. To further evaluate the potential safety concerns associated with systemic MR1 ligand stimulation, we will include a new experiment assessing the effects of MAIT cells plus 5-OP-RU in tumor-free animals. We will also discuss the potential specificity and safety limitations of systemic MR1 agonism in the revised manuscript.

Reviewer #2 (Public review):

The manuscript by Zhu et al. describes MAIT cell activation by riboflavin metabolites presented by MR1. The authors provide solid evidence for this activation and anti-cancer functional consequence using an array of selected cell lines, primary ex vivo and engineered xenograft models. Broadly, the results are thorough and well controlled, and provide a highly informative insight into the metabolite-MAIT-cancer cell interactions. However, the majority of this work is undertaken using models that preferentially express key targets, and whilst still useful, the (current) broader implications of this research are overstated. Additionally, the suggested MAIT modulation of the tumor microenvironment requires clarification.

We thank the reviewer for the positive feedback. We will address all comments and suggestions point by point.

Major Comments:

(1) In Figures 2b-d, the authors suggest microbial metabolite stimulation of PBMC cultures increased MAIT cell frequency up to 60%. Whilst their flow data is compelling, the frequency of one population can be influenced by changes in other populations. A form of absolute or relative-to-total count should be used.

We thank the reviewer for this valuable comment. We will provide absolute cell counts and/or normalized data to more accurately assess changes in MAIT cell frequency.

(2) The statements regarding cytokine induction in Figure 4e are too strong; many of those inflammatory cytokines are not automatically and consistently tumour-suppressive. The line 299 '...were not induced' may just reflect death of tumor cells. It would be useful to include tumour cell-only controls in Figure 4.

We thank the reviewer for this valuable comment. We agree that the statements regarding cytokine induction should be interpreted more cautiously. We will revise the relevant claims.

(3) Figure 7 is interesting, but the authors' conclusion that MAIT+5-OP-RU controls the tumor microenvironment is not robustly supported by their evidence.

(a) It is not clear how CD14+ cells established a sustained suppressive environment.

We thank the reviewer for this valuable comment. We will include additional experiments to further characterize the contribution of CD14+ cells to the observed suppressive environment.

(b) It is not clear how the peritoneal addition of microbial metabolites 'significantly enhanced MAIT-mediated tumor control'. The authors show that the addition of 5-OP-RU reduced the number of GFP-expressing tumour cells present in peritoneal lavage fluid. There is limited evidence to suggest this occurs through MAIT cells or MR1 in this figure.

We thank the reviewer for this valuable comment. We will include additional T-cell and T-cell + 5-OP-RU control groups to further determine the contribution of MAIT cells to the observed tumor control.

(c) It is difficult to draw conclusions from peritoneal lavage flow when some experimental groups received cells IP, but then all groups were equally assessed for key populations, and all data are presented as frequencies. The authors should use absolute counts (or similar) to appropriately show changes in cell populations to account for varying total/live/cd45+ cell compartments.

We thank the reviewer for this valuable comment. We will provide absolute cell counts, in addition to frequencies, to account for differences in total and viable CD45+ cell numbers.

(d) It would be necessary at a minimum to include 5-OP-RU-only controls, and ideally include MR1 blocking or the cancer line with MR1 removed. Alongside this, the authors should substantially reduce the strength of their statements on microbial metabolite-MAIT suppression of the tumor microenvironment.

We thank the reviewer for this valuable comment. We will include additional T-cell and T-cell + 5-OP-RU control groups and will substantially moderate our statements regarding microbial metabolite-mediated modulation of the tumor microenvironment.