A nanounit strategy reverses immune suppression of exosomal PD-L1 and is associated with enhanced ferroptosis

In addition to increasing the expression of programmed death-ligand 1 (PD-L1), tumor cells can also secrete exosomal PD-L1 to suppress T cell activity. Emerging evidence has revealed that exosomal PD-L1 resists immune checkpoint blockade, and may contribute to resistance to therapy. In this scenario, suppressing the secretion of tumor-derived exosomes may aid therapy. Here, we develop an assembly of exosome inhibitor (GW4869) and ferroptosis inducer (Fe3+) via amphiphilic hyaluronic acid. Cooperation between the two active components in the constructed nanounit induces an anti-tumor immunoresponse to B16F10 melanoma cells and stimulates cytotoxic T lymphocytes and immunological memory. The nanounit enhances the response to PD-L1 checkpoint blockade and may represent a therapeutic strategy for enhancing the response to this therapy.

1. The authors should provide clear evidence of cell death induction by ferroptosis in their in vivo models in the tumors as a result of their therapy. The authors measured SLC7A11, SLC3A2, GSH, lipid peroxidation and Cyctine but did not provide direct evidence and experimental proof of cell death induction. So, a direct experimental proof (e.g., either by TUNEL staining in combination with the markers mentioned above or using other methods) should be provided of the occurrence of ferroptotic cell death in vivo under the investigated experimental conditions. 2. Next, it is essential to exclude also the contribution of other cell death modalities, for example like apoptosis and necroptosis, which can be done either using IHC (active/cleaved caspase-3 staining and RIPK3P) or by inclusion in the study additional specific cell death inhibitors such as zVAD-fmk for apoptosis and Nec-1s for necroptosis.
Minor comments 1) Please check English translations everywhere. Some parts are not clearly written, and spelling errors are present. Please carefully proof-read the article. 2) Recently, the immunogenicity of ferroptosis has been recently described in the field of immunogenic cell death (PMID: 33188036, PMID: 30686534). The authors should provide a more complete up-to-date state of the art and reflect in their discussion in regard to already published data.
3) CD44 function is never explained, this is not clear for the reader why cells with CD44 overexpression are targeted. 4) 'Quantitative plunging of tumoral exosomes': this is not clear; please explain. 5) In the methods section, there is no explanation about cell culture. Please add. 6) Multicellular spheroids means that it consists out of multiple cell types. This is not the case here. 7) Figure 2f: no control for GW only is added to the graph. 8) Figure 4o: this figure is a repetition of figure 1b. 9) In figure 5f other populations can be seen in the gating. This is confusing, are those cells also seen as CD44+. Also, in this figure, you mention looking at CD44high, but only with high fluorescence intensity; you can say it is CD44high, not the whole range. Not everything can be high.
Reviewer #2: Remarks to the Author: In this paper, the authors report on nano-assembly of exosome inhibitor (GW4869) and ferroptosis inducer (Fe3+) in amphiphilic hyaluronic acid to evoke potent anti-tumor immuno response to hypermetastatic B16F10 melanoma. The results showed that combination of GW4869 and ferroptosis can significantly reduce exosomes and improve the immunotherapy for melanoma. The concept of this paper is interesting, and the paper is well-written. It can be accepted after following revisions. 1. Fig.3a: HGF induced obviously more effective exosome inhibition than GW4869. In comparison, downregulation of PD-L1 was not so great. The claimed effect of HGF on exosomal PD-L1 can not be supported by these data. The author needs to clarify this or change their claims. 2. The authors need to explain why HGF exhibited similar reduction of exosomes to HACA-GW. Given the fact that ferroptosis would cause cell death and accordingly reduction of exosomes, it is important to have a control with HACA-Fe on the exosomes. Furthermore, it would be more important to measure the exosomal PD-L1 in the tumor tissue. 3. Are there any connections between exosomes and PD-L1 expression on the tumor cells? It would be better if the authors could collect and analyze the PD-L1 expression on both B16-F10 cells and exosomes. 4. Since there were more memory T cells in spleen, it would be more interesting to see whether there is long-term tumor inhibitory effect. In general, the therapeutic experiments are too short.
Reviewer #3: Remarks to the Author: The authors developed nanoparticles (HGF) by combining ceremide inhibition drug GW4869 and ferroptosis inducer Fe3+. By suppressing T cell function via exosomal PD-L1 and Fe3+-induced ferroptosis in tumor cells, HGF was claimed to inhibit tumor growth and metastasis, enhance checkpoint blockade. Using amphiphilic HACA to encapsule the hydrophobic GW4869, and targeting CD44high tumor cells, the nanoparticles showed anti-tumor effect and induced T cells activation in vivo. However, there are also some a number questions in this manuscript: 1. Fig.3c, it would be hard to understand why there was no CD63 from enriched exosomes in the western blot figure. Also, how ferroptosis affects the level of PD-L1 should be addressed or at least discussed.
2. Fig.4a,b, the exosomes should be used instead of using the B16F10-cultural medium to study the effect of exosomes in this process. The media contain multiple factors and the drugs left in the medium might affect the results. Furthermore, it would be important to show the direct effect of these drugs on T cells.  Fig. 5F, need to provide the gating rationale or control for the CD44 high population.

5.
6. As HA on the nanoparticle can help to target CD44 overexpressed tumor cells, how about the side effect on activate T cells also with high level of CD44? 7. GW4869 influences a number of cellular activity related to the ceremide pathway on tumor cells, not just exosome secretion.
8. There is no direct evidence that the observed inhibitory effect is from PD-L1 on the exosomes. The statement on the role of exosomal PD-L1 in the title and the text is not warranted.
In the manuscript by Guohao Wang et al., the authors showed a novel combination therapy targeting both exosome secretion by cancer cells and ferroptosis. In their article, the authors developed a nanounit strategy that reverses immune suppression of exosomal PD-L1 and is associated with enhanced ferroptosis in vivo. This strategy is exciting, and it can increase immune response in therapy, especially when combined with PDL1 checkpoint blockade. Thus, this work is very interesting for translational research as well. This work is novel and original and may lead to improved anti-cancer therapy and unravel the role of ferroptosis in cancer therapy. The data support the main conclusions. However, there are several important issues which the authors should address before this article can be published in Nature Communications.
Major comments 1. The authors should provide clear evidence of cell death induction by ferroptosis in their in vivo models in the tumors as a result of their therapy. The authors measured SLC7A11, SLC3A2, GSH, lipid peroxidation and Cyctine but did not provide direct evidence and experimental proof of cell death induction. So, a direct experimental proof (e.g., either by TUNEL staining in combination with the markers mentioned above or using other methods) should be provided of the occurrence of ferroptotic cell death in vivo under the investigated experimental conditions. 2. Next, it is essential to exclude also the contribution of other cell death modalities, for example like apoptosis and necroptosis, which can be done either using IHC (active/cleaved caspase-3 staining and RIPK3P) or by inclusion in the study additional specific cell death inhibitors such as zVAD-fmk for apoptosis and Nec-1s for necroptosis.
Minor comments 1) Please check English translations everywhere. Some parts are not clearly written, and spelling errors are present. Please carefully proof-read the article. 2) Recently, the immunogenicity of ferroptosis has been recently described in the field of immunogenic cell death (PMID: 33188036, PMID: 30686534). The authors should provide a more complete up-to-date state of the art and reflect in their discussion in regard to already published data. 3) CD44 function is never explained, this is not clear for the reader why cells with CD44 overexpression are targeted. 4) 'Quantitative plunging of tumoral exosomes': this is not clear; please explain. 5) In the methods section, there is no explanation about cell culture. Please add. 6) Multicellular spheroids means that it consists out of multiple cell types. This is not the case here. 7) Figure 2f: no control for GW only is added to the graph. 8) Figure 4o: this figure is a repetition of figure 1b. 9) In figure 5f other populations can be seen in the gating. This is confusing, are those cells also seen as CD44+. Also, in this figure, you mention looking at CD44high, but only with high fluorescence intensity; you can say it is CD44high, not the whole range. Not everything can be high.
Reviewer #2 (Remarks to the Author): In this paper, the authors report on nano-assembly of exosome inhibitor (GW4869) and ferroptosis inducer (Fe3+) in amphiphilic hyaluronic acid to evoke potent anti-tumor immuno response to hypermetastatic B16F10 melanoma. The results showed that combination of GW4869 and ferroptosis can significantly reduce exosomes and improve the immunotherapy for melanoma. The concept of this paper is interesting, and the paper is well-written. It can be accepted after following revisions.
1. Fig.3a: HGF induced obviously more effective exosome inhibition than GW4869. In comparison, downregulation of PD-L1 was not so great. The claimed effect of HGF on exosomal PD-L1 can not be supported by these data. The author needs to clarify this or change their claims.
2. The authors need to explain why HGF exhibited similar reduction of exosomes to HACA-GW. Given the fact that ferroptosis would cause cell death and accordingly reduction of exosomes, it is important to have a control with HACA-Fe on the exosomes. Furthermore, it would be more important to measure the exosomal PD-L1 in the tumor tissue.
3. Are there any connections between exosomes and PD-L1 expression on the tumor cells? It would be better if the authors could collect and analyze the PD-L1 expression on both B16-F10 cells and exosomes. 4. Since there were more memory T cells in spleen, it would be more interesting to see whether there is long-term tumor inhibitory effect. In general, the therapeutic experiments are too short.
Reviewer #3 (Remarks to the Author): The authors developed nanoparticles (HGF) by combining ceremide inhibition drug GW4869 and ferroptosis inducer Fe3+. By suppressing T cell function via exosomal PD-L1 and Fe3+-induced ferroptosis in tumor cells, HGF was claimed to inhibit tumor growth and metastasis, enhance checkpoint blockade. Using amphiphilic HACA to encapsule the hydrophobic GW4869, and targeting CD44high tumor cells, the nanoparticles showed anti-tumor effect and induced T cells activation in vivo. However, there are also some a number questions in this manuscript: 1. Fig.3c, it would be hard to understand why there was no CD63 from enriched exosomes in the western blot figure. Also, how ferroptosis affects the level of PD-L1 should be addressed or at least discussed.
2. Fig.4a,b, the exosomes should be used instead of using the B16F10-cultural medium to study the effect of exosomes in this process. The media contain multiple factors and the drugs left in the medium might affect the results. Furthermore, it would be important to show the direct effect of these drugs on T cells. 5. Fig. 5F, need to provide the gating rationale or control for the CD44 high population.
6. As HA on the nanoparticle can help to target CD44 overexpressed tumor cells, how about the side effect on activate T cells also with high level of CD44? 7. GW4869 influences a number of cellular activity related to the ceremide pathway on tumor cells, not just exosome secretion.
8. There is no direct evidence that the observed inhibitory effect is from PD-L1 on the exosomes. The statement on the role of exosomal PD-L1 in the title and the text is not warranted.

REVIEWER COMMENTS
Note to reviewers: We really appreciate the reviewers to give insightful comments to further improve our manuscript. Here, the detailed point-by-point responses have been provided to address all the comments raised by the reviewers. and necroptosis inhibitor (necrostatin-1s, Nec) were adopted to co-culture with B16F10 cells, separately. From the cellular viabilities in the Supplementary Fig. 19 below, we noticed that HGF-induced cell death was rescued by the ferroptosis inhibitor (ferrostatin-1) only, but not by the apoptosis inhibitor (Z-VAD-FMK) or the necroptosis inhibitor (Nec), which is a direct experimental proof to demonstrate the main ferroptotic cell deaths in vivo.
Furthermore, the levels of apoptosis and necroptosis in tumor tissues of different groups were detected via an immunofluorescence staining method by using apoptosis marker (cleaved caspase-3) and necroptosis marker (RIPK3P), respectively. No stronger fluorescence signal of the cleaved caspase-3 (Supplementary Fig. 25) or RIPK3P (Supplementary Fig. 26) was observed in HGF NPs group compared to the other three groups. Together, typical ferroptosis played the key role in the therapeutic function of HGF group.

3) CD44 function is never explained, this is not clear for the reader why cells
with CD44 overexpression are targeted. The corresponding context has been added in line 71-73 of page 3, in the revised manuscript.

4) 'Quantitative plunging of tumoral exosomes': this is not clear; please explain.
Response: Herein, 'quantitative plunging' was adopted to indicate the significant elimination of tumoral exosomes, due to the functional HGF. This inappropriate vocabulary has been corrected as 'Significant elimination of tumoral exosome' in line 145 of page 6, in our revised manuscript.

5) In the methods section, there is no explanation about cell culture. Please add.
Response: The explanation has been added in the 'Cell lines and culture media' section in the 'methods' part of our revised manuscript.

6) Multicellular spheroids means that it consists out of multiple cell types. This is not the case here.
Response: The inappropriate vocabulary has been corrected as 'B16F10 cancer cellular spheroids' line 386 of page 14, in our revised manuscript.

7) Figure 2f: no control for GW only is added to the graph.
Response: As this reviewer kindly suggested, absorption of GW4869 has been examined and added into Fig. 2f in our revised manuscript.  The author needs to clarify this or change their claims.
Response: Compared with PBS and GW4869 groups, we observed that HGF had the most significant inhibition on tumoral exosomes, demonstrated by the intensity of exosome biomarker CD63. However, the gray intensity of PD-L1 of the HGF group was higher than that of the CD63, which might be due to that the content of exosomal PD-L1 was higher than the content of CD63 on the secreted exosome. Overdose of total protein thus cannot highlight the PD-L1 variation very well.
To make the results more convincing, we further optimized the experimental protocol, via reducing the amount of protein loaded and using PD-L1 antibody with higher sensitivity, to repeat the experiment and confirm the inhibition effect of HGF on exosomal PD-L1. As anticipated, higher intensity of PD-L1 was found in each group, compared to the intensity of CD63. GW4869 from HGF NPs diminished the exosome biomarker CD63 of B16F10 cells and exosomal PD-L1 dramatically. Fig. 3a has been corrected in our revised manuscript.

Fig. 3a
Western blot analysis for exosome marker CD63 and PD-L1 in the medium of B16F10 cells after treatment. GAPDH served as a loading control.
Images were representative of three experiments.

The authors need to explain why HGF exhibited similar reduction of exosomes to HACA-GW. Given the fact that ferroptosis would cause cell death and accordingly reduction of exosomes, it is important to have a control with
HACA-Fe on the exosomes. Furthermore, it would be more important to measure the exosomal PD-L1 in the tumor tissue.
Response: From our previous western blot figure, CD63 intensities in HACA-GW and HGF groups may look similar by our naked eyes. However, obvious quantitative differences can be achieved by using the Image J software, as shown in the Explanatory figure 1.

Explanatory figure 1. CD63 protein level of different groups.
Additionally, overdose of total protein in the western blot trial may cloud CD63 differences in HACA-GW and HGF groups further. Therefore, in our new trail with an added HACA-Fe group, loading amount of total protein was optimized (Fig. 3c in the revised manuscript). Lowest level of CD63 detected in HGF group convinced the effective exosome reduction via the combination of GW4869 and enhanced ferroptosis. To further analyze the effect of HACA-Fe on exosome secretion in vitro (Fig. 3a) and in vivo (Fig. 3c) through the western blot assay, we noticed that the secretion of exosomes decreased slightly, which may be explained by cellular ferroptosis.
HACA-GW reduced the exosomal PD-L1 effectively. A combination with ferroptosis inhibited the exosomal PD-L1 in HGF group further.

PD-L1 expression on both B16-F10 cells and exosomes.
Response: Thanks for the reviewer's insightful question. Recent studies showed that exosomal PD-L1 has the same membrane topology as cell surface PD-L1, with its extracellular domain exposed on the surface of the exosomes. A similar PD-L1 level exists in mouse metastatic melanoma B16F10 cells and purified exosomes (Chen, G., Response: Long-term tumor inhibitory effect and survival curves had been included in the Fig. 5b, c, d and Fig. 6b, c, d in the revised manuscript. The corresponding context has been corrected in line 220-226 of page 8-9 and line 241-246 of page 9, in the revised manuscript. n=5. Data was presented as mean ± s.d. The analysis method was One-way ANOVA with Tukey's post hoc test. Significance was presented as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. Response: Exosomes in Fig. 3c were isolated from tumor tissues with the same weight after different treatments. Since HACA-GW and HGF NPs could inhibit the exosome secretion from the tumor site significantly, low level of exosome biomarker CD63 in these two groups were therefore collected. To analyze the effect of ferroptosis on exosomal PD-L1, four groups including PBS, HACA-Fe, HACA-GW and HGF were prepared, in which HACA-Fe could cause ferroptosis individually. Ferroptotic cell death in HACA-Fe group reduced the total secretion of exosomes sourcing from cancer cells (Fig. 3a and 3c).   2. Fig.4a, b, the exosomes should be used instead of using the B16F10-cultural medium to study the effect of exosomes in this process. The media contain multiple factors and the drugs left in the medium might affect the results.
Furthermore, it would be important to show the direct effect of these drugs on T

cells.
Response: It is true that many factors in the cell culture medium that may affect the results, and that is why the total exosome isolation reagent (Thermo Fisher, Cat. No. 4478359) was applied in Fig. 4a to collect the exosomes first. To avoid the misunderstanding, the experimental details in the method section has been perfected as: 'Total exosome isolation reagent (Thermo Fisher,Cat No. 4478359) was applied to collect the secreted exosomes. In detail, the reagent was added to the cell supernatant for overnight culturing at 2 °C to 8 °C to precipitate the exosomes out.
The precipitated exosomes were then achieved by standard centrifugation at 10,000 g for 60 min. Co-culturing the collected exosomes and CD8 + T cells can detect the exosomal effect on T cells.' As this reviewer suggested, the direct effect of different drugs on T cells has been measured. Supplementary Fig. 3a indicated that HACA-GW had no effect on the release of IFN-γ in T cells comparing with PBS. However, level of IFN-γ decreased in HGF and HACA-Fe groups by 9.7% and 8.6%, respectively. We also tested the cell viability of T cells after incubation with PBS, HACA-Fe, HACA-GW and HGF NPs.
At the highest concentration of HACA-Fe and HGF NPs, the viability of T cells was slightly inhibited and remained above 80% (81.2% at 100 µM for 48 h) ( Supplementary Fig. 3b. in our revised manuscript). Thus, the inhibition of IFN-γ in these two groups might be due to the decreased cell viability induced by Fe 3+ -relevant ferroptosis. and HACA-Fe groups (Fig. 4m). A corresponding decreasing of GPX4 activity was thus triggered (Fig. 4e, n) via using tert-butylhydroperoxide (tBuOOH) as a substrate and monitoring the rate of NADPH oxidation. However, treating T cells with HGFs increased GPX4 activity surprisingly ( Supplementary Fig. 22). Such phenomena may be due to the upregulated cystine and cysteine transporter activities in activated T cells, increasing the GSH synthesis and GPX4 activity (Ishii, T., et al. J Cell Physiol 1987, 133, 330-336;Garg, S. K., et al. Antioxidants & Redox Signaling 2010, 15, 39-47;Levring, T. B., et al. Sci Rep 2012, 2, 266). Collectively, HGF NPs upregulated the level of GPX4 activity in T cells, but downregulated whose function in tumor cells conversely, by virtue of the reduction of exosomes and the enhanced ferroptosis ( Fig. 3d-f). Different roles in T cells and tumor cells may associate with the robust activation of CD8 + and CD4 + T in our nanounit system. presented as mean ± s.d. The analysis method was One-way ANOVA with Tukey's post hoc test. Significance was presented as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. n=5. Data was presented as mean ± s.d. The analysis method was One-way ANOVA with Tukey's post hoc test. Significance was presented as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. One-way ANOVA with Tukey's post hoc test. Significance was presented as *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.