Bioengineered bacteria-derived outer membrane vesicles as a versatile antigen display platform for tumor vaccination via Plug-and-Display technology

An effective tumor vaccine vector that can rapidly display neoantigens is urgently needed. Outer membrane vesicles (OMVs) can strongly activate the innate immune system and are qualified as immunoadjuvants. Here, we describe a versatile OMV-based vaccine platform to elicit a specific anti-tumor immune response via specifically presenting antigens onto OMV surface. We first display tumor antigens on the OMVs surface by fusing with ClyA protein, and then simplify the antigen display process by employing a Plug-and-Display system comprising the tag/catcher protein pairs. OMVs decorated with different protein catchers can simultaneously display multiple, distinct tumor antigens to elicit a synergistic antitumour immune response. In addition, the bioengineered OMVs loaded with different tumor antigens can abrogate lung melanoma metastasis and inhibit subcutaneous colorectal cancer growth. The ability of the bioengineered OMV-based platform to rapidly and simultaneously display antigens may facilitate the development of these agents for personalized tumour vaccines.


Main Points
1) The tumor models used in this paper are simplistic and do not robustly test the efficacy of the system. TRP and OVA antigens are highly immunogenic peptides that their efficacy has been showed in combination with an appropriate adjuvant. I would like to see an experiment where this system has been tested with different neo-antigens and tumor models other than the B16 family. In addition the lung metastasis model is interesting but survival curves and tumor growth would also be needed to complete the study.
2) A more in-depth immunological analysis would be needed to evaluate what kind of immunity the OMV are eliciting, only T cells? Only CD8 T cells? What kind of phenotype these T cells show? Are they all effector? What about memory in comparison with more classical vaccine approach (Poly-IC + antigen for example).
3) I believe this system works really well because of the ability of the OMVs to travel to lymph nodes efficiently hence explaining why OMV conjugates with the peptides resulted better than the mixture of them. However, to fully elucidate the mode of action it would be necessary to see a group of mice treated with Poly IC and the peptides (or other adjuvants).
4) The vaccination effect of this system was not very thoroughly tested. Two types of experiments could have been done. First type of experiment would be vaccinating the mice with the OMV system and then testing the engraftment of the tumor. Second type of experiment would be rechallenging the mice with the same tumor. Abscopal effect could have been also a useful model to shed light on the mechanism. 3) In the first mice experiment with the OVA-model, an increase of CD4+ T cells is shown but this could be an increase of Tregs to counterbalance the increased immunity. Authors should examine that this increase is not a T-reg increase but rather a Th1 response (since it's a bacterial component) 4) In Supplementary figure 10, increase the FBS percentage to at least 50% to show how robust this method is and imitating in vivo setting as close as possible 5) Supplementary figure 13, in the CD80 and CD86+ upregulation experiment it would be beneficial to add also CN-OMVs alone and CO-OVA just to check whether the linkage of this SpT/SpC pair and SnT/SnC pair affect DC maturation 6) In Figure 3G why is the SpT-OVA-Cy5.5 condition showing more presentation compare to the CC-SpT-OVA-Cy5.5 OMVs? please discuss 7) In Figure 3h, why were the formulations injected intradermal?? Wouldn't it be better to do SC as done previously in the efficacy tests? please discuss 8) In Fig4g, why is there more inf-gamma secreted with CN OMVs compared to SnT-TRP2? It  seems like the OMVs have TRP2 peptides or other peptides with high homology? 1. The ClyA fusion proteins are said to be surface-exposed, but I don't see the evidence for that. Only localization to the OMVs is demonstrated, not surface localization as with the plug-and-display system. Response 1: Thank you very much for the comment. The site of the fusion protein should avoid affecting the protein structure as much as possible. As a transmembrane protein, the suitable modification sites of ClyA are N-terminal and C-terminal. In this study, we fused the catchers to the C-terminal of ClyA, which is common in the previous studies 1-3 . We confirmed the location of catchers through their reaction with the gold nanoparticles-labelled tags (Figure 3e).
We also tried to fuse the catchers into the N-terminal of ClyA. According to the structure of ClyA ( Figure R1 and R2) 4,5 , both N-terminal and C-terminal of ClyA are extracellular. As shown in Figure R3, either N-terminal or C-terminal fusion with SpyCatcher on ClyA can bind SpyTag, indicated by the bands at the molecular weight of about 45 kDa.    Response 2: The E.coli strain used in the study is Rosetta (DE3), which is the popular stain in genetic engineering and has wildtype LPS expression. We detected that the level of LPS in the OMVs (1 mg/mL) was 120 ng/ml. Compared with the half lethal dose (LD 50 ) of LPS (300 μg/mouse), the dose of OMVs (50 μg OMV per mouse, 6 ng LPS per mouse) we used in this study is significantly lower. We also evaluated the effects of OMVs on cell activity in vitro. The data show that OMVs were biocompatible and non-toxic at all concentrations tested when incubated with BMDCs cells in vitro (new Supplementary Figure 3). The biosafety of the OMVs has also been confirmed in clinic. The OMVs-based Group B meningococcal vaccine MeNZB has effectively limited the incidence and mortality of meningitis in New Zealand.

Responses to Reviewer 4
This MS of Cheng...Nie presents a new approach to vaccination using OMV nanoparticles of GN bacteria and a ClyA-fusion system for capture of one or more antigens for presentation. The authors suggest both induction of innate and adaptive immune responses in a murine test system, and support this with data. They do not adequately describe the ClyA system and this ought to be addressed. While their focus is upon developing an approach to tumor antigens, and specifically neoantigens, the question arises at this time due to pandemic regarding suitability of this approach to SARS CoV2. It would be of interest to see studies upon human solid tumors to address the clinical question that confronts oncology, but this clearly lies outside the domain of the authors' work.
Thank you very much for the positive comment. Cytolysin A (ClyA) is a pore-forming toxin synthesized by Escherichia coli and other enteric bacteria. ClyA has been used to display microbial antigens or model antigens in the previous studies 6,7 . As a transmembrane protein, the suitable modification sites of ClyA are N-terminal and C-terminal. In this study, we fused the catchers to the C-terminal of ClyA, which is commonly used in the previous studies for the surface modification of OMVs. In this study, the OMVs-based vaccine platform was mainly used to induce T-cell-mediated immunity. Several studies showed that T-cell-mediated immunity plays an important role in COVID-19 prevention and treatment 8-10 .

A smaller issue is whether functional effects of F/T were assessed beyond the morphological effects reported L211.
Response 1: Thank you for the helpful comment. We have assessed the function of frozen-thawing CC-SpT-OVA OMVs in the revised manuscript. After 12 h of co-culture with different formulations (PBS, fresh CC-SpT-OVA OMVs or repetitive freezing-thawing CC-SpT-OVA OMVs), BMDCs were collected for flow cytometry analysis for maturation (CD11c + CD80 + CD86 + ) and antigen presentation (CD11c + MHC I-OVA + ). The results were shown in the new Supplementary Figure  10d-e, which was added into manuscript as follow: After conjugating with SpT-OVA, repetitive freezing-thawing CC-SpT-OVA OMVs could also effectively stimulate the maturation of BMDCs (Supplementary Figure  10d) and promote the presentation of antigens (Supplementary Figure 10e), with no significant difference compare to the non-freezing-thawing (fresh) preparation. Figure 10.

There are multiple typographical and syntactic issues (article 'the' omitted L32) and innate composition (that has no immediate interpretation for this reviewer L64). There are multiple areas where the symbol⏍ 'box' is inserted in place of ?'C' L 119 and 'ff' L147, L546)
Response 2: Thanks for your help. We have carefully read our manuscript and improved our description accordingly.

Main Points
1) The tumor models used in this paper are simplistic and do not robustly test the efficacy of the system. TRP and OVA antigens are highly immunogenic peptides that their efficacy has been showed in combination with an appropriate adjuvant. I would like to see an experiment where this system has been tested with different neo-antigens and tumor models other than the B16 family. In addition the lung metastasis model is interesting but survival curves and tumor growth would also be needed to complete the study.
Response 1: Thank you very much for the constructive comment. According to your suggestion, we further used the OMVs-based vaccine platform to display and deliver a real tumor neoantigen, Adpgk, and evaluated the anti-tumor immunity in the subcutaneous MC38 tumor model. The results were shown in new Figure 7, and were added into the manuscript as follows:
We systemically studied the immune memory, and the results were added into the revised manuscript as follows:

The long-term immune memory in vivo elicited by antigen-loaded CC OMVs
Successful induction of immune memory is critical for long-term benefit of tumor vaccine. For immune memory studies, the mice were vaccinated on days 0, 3 and 8 (Figure 8a) Figure 22a, the splenocytes exhibited a greater cytotoxic effect against B16-OVA cells in CC-SpT-OVA OMVs group than that in other groups. This effect disappeared in the experiments using MC38 cells without OVA antigen, which provide robust evidence for the antigen specificity of immune response (Figure 8d  and Supplementary Figure 22b). Next, we quantified antigen-specific T cells in splenocytes and blood by flow cytometry. CC-SpT-OVA OMVs vaccination elicited more antigen-specific T cells (tetramer + T cells) and IFNγ + cytotoxic T lymphocytes than Poly (I:C) + SpT-OVA and SpT-OVA + CN OMVs, respectively (Figure 8e-8g  and Supplementary Figure 23a, 23b and 24). Furthermore, CC-SpT-OVA OMVs induced obvious central memory T cell (~24%) and effector memory T cells (~7%) (Figure 8h and Supplementary Figure 25a, 25b) for over 60 days, indicating that CC-SpT-OVA OMVs could be used as prophylactic vaccine. On days 60, the immunized mice were challenged with i.v. injection of 2 × 10 5 B16-OVA cells. In contrast to the obvious lung metastasis in the mice immunized with mixture formulations, there was almost no lung metastasis in mice in CC-SpT-OVA OMVs group (Figure 8i and 8j).
3) I believe this system works really well because of the ability of the OMVs to travel to lymph nodes efficiently hence explaining why OMV conjugates with the peptides resulted better than the mixture of them. However, to fully elucidate the mode of action it would be necessary to see a group of mice treated with Poly IC and the peptides (or other adjuvants).

Response 3.
Thank you for your helpful suggestion. We have added two animal experiments to compare the anti-tumor efficacy of our vaccine platform and mixed formulation (SpT-antigen + Poly (I:C)), which consists of approved adjuvants and antigen. Compared to the Poly (I:C), the OMVs-based platform induced stronger anti-tumor immunity in the subcutaneous MC38 tumor model (new Figure 7) and stimulated more immune memory cells to protect the mice from following tumor cells challenge (new Figure 8).

4)
The vaccination effect of this system was not very thoroughly tested. Two types of experiments could have been done. First type of experiment would be vaccinating the mice with the OMV system and then testing the engraftment of the tumor. Second type of experiment would be re-challenging the mice with the same tumor. Abscopal effect could have been also a useful model to shed light on the mechanism. Response 4. Thank you for the insightful comment. We have completed these two types of experiments, which were shown as new Figure 8. Vaccination using our OMVs-based platform stimulated significant immune memory and protected the mice from the following tumor cells challenge. In addition, 50% vaccine-cured mice exhibited complete tumor resistance against the tumor cells re-challenge. As expected, the infiltration of CD3 + T cells, CD3 + CD8 + T cells, CD3 + CD4 + T cells, activated neutrophils (CD11b + Ly6G + cells) and DCs (CD11c + cells) were all significantly elevated in MC38 tumor tissues after s.c. immunization with the CC-SpT-Adpgk OMVs (Figure 7e and Supplementary Figure 21). The immunosuppressive microenvironment mediated by regulatory T cells (Treg, CD3 + CD4 + Foxp + T cells) was alleviated effectively by CC-SpT-Adpgk OMVs treatment (Figure 7e). These immunomodulation effects in CC-SpT-Adpgk OMVs group were more dramatic than that in the mixture groups. There was no significant change in the infiltration of macrophages (F4/80 + cells) in the tumor tissue between different groups. Interestingly, there was an infiltration of myeloid-derived suppressor cells (MDSCs, CD11b + Gr + cells) after CC-SpT-Adpgk OMVs treatment, although the MDSCs infiltration did not disturb the anti-tumor effect (Figure 7e).

6) Why the authors are using only ELISPOT (please indicate on the y axes the amount of splenocytes tested, this info is only in material and methods) to check the presence of T cells specific response? TRP2 or SIINFEKL tetramer do exist.
Response 6. Thank you for your helpful suggestion. We have included the amount of splenocytes tested on Y axes in Figure 2n, 4h, 5d and 6d, according to the suggestions. We also supplemented SIINFEKL tetramer detection experiment in the revised manuscript (new Figure 8e and 8f).

Minor Comments 1) Supplementary figure 3, cell viability assay with different methods would have been beneficial. What about 7AAD and Annexin-V to further show this.
Response 1: We have optimized the experiment in the revised manuscript as follows: We first confirmed that OMVs were not toxic to murine BMDCs in the concentration range used in the current study, using annexin V-APC/7-AAD apoptosis detection assay to stain dead cells (Supplementary Figure 3). Figure 3. The cytotoxicity of OMVs in murine bone marrow-derived dendritic cells (BMDCs) was measured by flow cytometry after 24 h incubation with CO OMVs at the indicated protein concentrations or PBS. Cells were stained with annexin V-APC/7-AAD. The Annexin V -/7-AADcells are viable cell.

New Supplementary
2) Line 147, there are two f's missing in effect.

Response 2:
We have revised the sentences and carefully checked the whole manuscript.
3) In the first mice experiment with the OVA-model, an increase of CD4 + T cells is shown but this could be an increase of Tregs to counterbalance the increased immunity. Authors should examine that this increase is not a T-reg increase but rather a Th1 response (since it's a bacterial component).

Response 3:
We greatly appreciate the constructive suggestions made by this reviewer. As shown in new Figure 7e, we found that the infiltration of CD3 + CD4 + T cells were elevated and regulatory T cells (Treg, CD3 + CD4 + Foxp + T cells) were alleviated effectively by the CC-SpT-Adpgk OMVs treatment, indicating CC-antigens OMVs alleviate the tumor immunosuppressive microenvironment. Supplementary figure 10, increase the FBS percentage to at least 50% to show how robust this method is and imitating in vivo setting as close as possible.

Response 4:
According to the suggestion, CC OMVs were incubated with 50% fetal bovine serum (FBS) and the morphology was characterized by TEM. As the results shown in Figure R4, the background of TEM image is dirty due to high content of protein in PBS containing 50% FBS. Importantly, we still found that the morphology of CC OMVs was unaffected after 24 h incubation, suggesting that CC OMVs are likely to remain stable enough for vaccination to be effective.  Figure 2a and 2b, we measured the proportion of CD80 + and CD86 + in CD11c + BMDCs cultured with CN OMVs, OVA 257-264 + CN OMVs or CO OMVs, and found that all three formulations induced a significant increase in the proportion of CD80 + and CD86 + . Figure 3G why is the SpT-OVA-Cy5.5 condition showing more presentation compare to the CC-SpT-OVA-Cy5.5 OMVs? please discuss. Figure 3g showed the cell uptake results. The antigen presentation was shown in Supplementary Figure 13c. There was no significant difference in the amount of MHCI-OVA complex between SpT-OVA and CC-SpT-OVA OMVs. We hypothesize that OMV may promote the escape of the endosomes due to membrane fusion, thus enhancing cross-presentation. Figure 3h, why were the formulations injected intradermal?? Wouldn't it be better to do SC as done previously in the efficacy tests? please discuss.

Response 7.
Thank you for pointing out this mistake. Animals used in this study were all immunized by subcutaneous injection into the tail base. We have revised the description and checked the whole manuscript.

8) In Fig4g, why is there more inf-gamma secreted with CN OMVs compared to
SnT-TRP2? It seems like the OMVs have TRP2 peptides or other peptides with high homology? Response 8. It has been validated that OMVs suppress tumor growth by interferon-γ-mediated anti-tumor response 11 . The killed tumor cells will release tumor antigens, including TRP2 for recognition by the immune system, resulting in an increase in IFNγ + cytotoxic T lymphocytes. In spite of the presence of the specific antigen in SnT-TRP2 group, the lack of adjuvant effect of OMV to stimulate innate immunity may lead to an inability to stimulate effective antigen specific immune response.