Endogenous stimulus-powered antibiotic release from nanoreactors for a combination therapy of bacterial infections

The use of an endogenous stimulus instead of external trigger has an advantage for targeted and controlled release in drug delivery. Here, we report on cascade nanoreactors for bacterial toxin-triggered antibiotic release by wrapping calcium peroxide (CaO2) and antibiotic in a eutectic mixture of two fatty acids and a liposome coating. When encountering pathogenic bacteria in vivo these nanoreactors capture the toxins, without compromising their structural integrity, and the toxins form pores. Water enters the nanoreactors through the pores to react with CaO2 and produce hydrogen peroxide which decomposes to oxygen and drives antibiotic release. The bound toxins reduce the toxicity and also stimulate the body’s immune response. This works to improve the therapeutic effect in bacterially infected mice. This strategy provides a Domino Effect approach for treating infections caused by bacteria that secrete pore-forming toxins.

How to solve multi-drug resistant bacterial infections is a serious problem at present. The current manuscript investigates the influence of the targeting of bacterial toxins by nanomachines coated with calcium peroxide and rifampicin and the mechanism by which gas-triggered nanomachines accelerate the release of antibiotics. This antibacterial strategy is absolutely new and quite interesting. Reviewer recommends the paper for publication in Nature Communications after addressing the following minor issues. Details are as following： 1.There are many small mistakes in the article. For example, in line 191 of page 8, "d to evaluate" should be "du". Please carefully check and correct them. 2.In Figure 1e, the different absorption intensity of RFP in the groups of Free RFP (black), RFP in ethanol (red), in toxin (blue) should be explained and the corresponding content of RFP should be provided. 3.In Figure 2j, why did the RFP@PCM@Lec+toxin cannot release RFP at 37 oC? Is the size of the RFP smaller than the pore induced by toxin? How about the structure of nanomachines at 37 oC for 150 min? 4.The clarity of the pictures in the article is not enough, the resolution of the pictures should be improved. Some pictures are wrong, for example, the plates in figure 3a are arranged differently (In Figure 3a, some images of group I and group II are placed repeatedly, please check out and correct it), and in figure 4, the descriptions of b and c are opposite. 5.The format of the references should be checked. 6.If possible, repeatability measurements (error bars) should be included in all episodes, and they need to put their overall impact results in the background. For example, there has been a significant reduction in the number of MRSAs after treatment, but they do not provide a reference to compare their results with current studies of current clinical MRSA treatments or new antibacterial treatments for MRSA. 7.How to prove that the calcium peroxide nanoparticles are coated inside the lecithin particles, rather than having some adsorption on their surface. 8.Obviously you should cite some of the latest antibacterial studies, such as the study of photothermal and photodynamic synergies.
Reviewer #3 (Remarks to the Author): Comment: In this manuscript, Han, et al. fabricate lipid (lauric acid and stearic acid) nanoparticles loaded with calcium peroxide (CaO2) and rifampicin (RFM) as antibiotic and coated with liposome, comprising of lecithin and DSPE-PEG. The authors claim that this nanoparticle can absorb and neutralize poreforming toxin, resulting in triggered release of antibiotic. Then this toxin absorbed nanoparticle can also elevate immune response against toxin. This work builds on two or more previously published works from Zhang's lab and Xia's lab. (shown in reference). The novelty is from using antibiotic triggered release in combination of vaccination effect of the same nanoparticles. Here, Han and colleagues present a good platform for combination treatment of bacterial infection and have done several experiments to proof their claims. However, there are some inconsistencies in the experiments and several important issues that are not discussed. As a result, this reviewer does not recommend this manuscript for acceptance in Nature Communication at this time.
Additional comments: 1. In results and discussion on evaluation of immune effect of nanomachines (line 209-221) and caption of figure 5 (line 403-409), approximately 80% of this paragraph and caption is exactly the same as previously published article. This considers as plagiarism and it is unacceptable, especially at Nature Communication's standard. the use of RFP to rapidly treat bacterial infection with high efficacy and reduce toxic side effect, maintaining a therapeutically effective drug concentration in systemic circulation for a longer period of time, thus potentially increasing patient compliance (J. Cell Sci., 2013, 126, 3043; ACS Appl. Mater. Interfaces, 2014, 6, 16895; Med. Sci. Monitor, 2018, 24, 473).

"In addition, why rifampicin is chosen should be clearly defined."
Response: In our study, RFP was chosen as the antibacterial drug against the Methicillin-resistant Staphylococcus aureus (MRSA) for the following reasons: (1) RFP was primarily a frontline drug for the treatment of bacterial infection and showed good antibacterial activity in treating MRSA-related infections (Proc. Am. Thorac. Soc., 2004, 1, 338;Lancet, 2001, 357, 40 ); (2) In our previous research, RFP was loaded into the metal organic framework, and the combination of UV-light (365 nm), pH triggered precise RFP release and zinc ions enables the light-activated nanocomposite to significantly inhibit MRSA-induced wound infection and accelerate wound healing (Adv. Funct. Mater., 2018, 28, 1800011); (3) We used vancomycin, a common antibacterial agent for MRSA infection, as control, and the drug release efficiency in nanoreactors were evaluated. The results are shown in Figure 1, and it can be seen that the nanoreactors for RFP loading has better drug release ability.
As previously mentioned, we used RFP as the antibacterial agent in our study. Comments-2: "The proposed nanoformulation hardly qualifies as &#x201C;nanomachine&#x201D; as claimed by the authors. Nanomachines typically are typically associated with mechanical movements, which are not described in the article at all. " Response: Thank you the good suggestion. Micro/nanoscale machines are a few micrometers to sub-micrometers scale devices that can harness power from various energy sources to generate mechanical motion in a controlled manner (Adv. Funct. Mater., 2018, 28(25): 1705867).
In our study, Calcium peroxide (CaO2) and rifampicin (RFP) are added into the lauric acid (LA) and stearic acid (SA) eutectic mixture to form phase change materials-based nanoformulations, then the DSPE-PEGylated-lecithin (Lec) is used to coat the nanoformulations as a gate material for fabricating toxin-responsive nanoparticles for drug release. Once encountering pathogenic bacteria in vivo, the nanoparticles are pierced by the toxins secreted by the bacteria to form pores, and a series of chemical reactions take place inside the nanoparticles to cause controlled release of the drug (Please see: Equation 1 and 2).
Therefore, we believe that the term nanoreactor is more suitable than nanomachine to illustrate our experiments based on these experimental phenomena and results. In the revised manuscript, we have replaced nanomachines with nanoreactors, and all the modified portions are highlighted in yellow. CaO2+H2O=Ca(OH)2+H2O2 (1) H2O2=H2O+O2 (2)

Comments-3: "The term, phase change material (PCM) is used repeatedly in the article, yet its significance is hardly discussed. Why is a phase change material necessary for the nanoformulation preparation? Can it not be prepared by simple liposomes?"
Response: Thanks for the good suggestion. In the revised manuscripts, we have added the significance of the phase change material (PCM) (please see: P4, Lines 9-11).
In recent years, a new type of functional material, phase change material (PCM), has been found to be able to quickly respond to temperature and transform into a transparent liquid phase for a controllable release of drugs (Angew. Chem. Int. Ed., 2014, 53, 3780; Adv. Mater., 2017, 29,1703702).
In our study, rifampicin (RFP) and calcium peroxide (CaO2) were wrapped inside a phase-change material (PCM) which is made of a eutectic mixture of naturally occurring fatty acids with a well-defined melting point at 35.2°C-38.3°C (Please see: Table S1 in the newly revised manuscript), then the DSPE-PEG modified lecithin(Lec) is used to encapsulate the PCM to form a stable liposome, and because of the higher melting point of lecithin, the inner PCM could dissolve into a liquid state at temperature close to that of human bodies, and the melted PCM could be protected from leakage by the phospholipid layer, which can show we used the PCM rather than the simple liposomes.
Our experimental results showed that the nanoparticles exist in the solid state thus preventing the payloads from leaking out through diffusion at a temperature below the eutectic point. However, when the local temperature is increased beyond the eutectic point, the nanoparticles will melt, leading to a quick release of payloads (Please see: Figure S6 and Figure 2j in the newly revised manuscripts).
In this system, the optimized ratio of DSPE-PEGylated-lecithin (3:1) is used to coat the eutectic mixture of two fatty acids as a gate material for fabricating toxin-responsive nanoparticles for drug release.

Figure 2
The optimization step to minimize PEG density to facilitate toxin interaction. The capture efficiency of different-ratio nanoformulations for the toxin (a, b) and the hemolysis ratio of different-ratio nanoformulations (c and d).
In the newly revised manuscript, we have added the antibacterial assays using the RFP-CaO2@PCM@Lec nanoreactors, RFP and CaO2, respectively. The results are shown in Figure 3 (Please see: Figure S8 in the newly revised manuscript). It can be seen that the H 2O2 from the CaO2 can only inhibit 62% of bacteria; however, when RFP and CaO2 were wrapped to form the nanoreactors, the antibacterial activity had a significant increase of ~100%.

Comments-6: "Antibacterial activity should be expressed in logarithmic scale rather than linear scale. An effective antibiotic drug is typically associated with 3-log reduction in bacterial load. Antibacterial activity should be expressed in logarithmic scale rather than linear scale. An effective antibiotic drug is typically associated with 3-log reduction in bacterial load."
Response: Thanks for the good suggestion. In the newly revised manuscript, we have provided the logarithmic scale for antibacterial activity (please see: Figure 3e and 3f). The results are shown in Figure 4, and it can be seen that the RFP-CaO2@PCM@Lec nanoreactors has 3.02 log reduction in bacterial load.

Comments-7: "H&E and Tunnel assays in figure 4d,e are hardly quantitative and don&#x2019;t depict how the absorbed toxin is safer than toxin. "
Response: Thanks for the good suggestion. In this revised manuscript, we have added four assays to confirm that our synthesized nanoreactors have the ability to capture toxins, and the absorbed toxin is safer that toxin.
(1) Hemolysis ratio assay was used to evaluate the absorbed toxin and pure toxin. For detailed experimental methods, please see: P16, Lines6-20; The results have been provided in the newly revised manuscript (Please see: Figure 4c and 4d ) and shown in Figure 5. It can be seen that the nano-toxin (absorbed toxin) has a lower hemolysis ratio than the pure toxin; (2) Skin lesion assays were used to evaluate the demonstrable oedema and inflammation for different material treatments. The results have been provided in the newly revised manuscript (Please see: Figure S11 ) and shown in Figure 6. It can be seen that the nano-toxin (absorbed toxin) has weaker skin lesions than the pure toxin. (3) Haematoxylin and eosin stained histological sections and blood routine examination assays were used to evaluate the inflammation with inflammatory infiltrate. For detailed experimental methods, please see: P17, Lines 18-22. The results have been provided in the newly revised manuscript (Please see: Figure 4f and S12 ) and shown in Figure 7. It can be seen that the nano-toxin (absorbed toxin) has weaker inflammation than the pure toxin ( Figure 7a). Meanwhile, the blood routine examination assays showed that the pure toxin leads to higher white blood cell (WBC) and Granulocyte (Gran), and an increase in these indicators means that the pure toxin induces stronger inflammation and body damage, while the nano-toxin did not produce significant inflammation and damage. (4) TUNEL staining assay was used to reveal the widespread apoptosis throughout. For detailed experimental methods, please see: P 17, Lines 10-17; the results have been provided in the newly revised manuscript (Please see: Figure 4e and S10) and shown in Figure 8. It can be seen that there was no obvious skin damage in the other four treatments, while the pure toxin displayed significant toxicity in vivo and induced stronger cell apoptosis. Combining the above results, our experiments show that absorbed toxin is safer than toxin. Figure 7 was completed in 10 days and yet the antibodies were generated after 21 days. The value of the vaccination arm is therefore questionable in the present study. " Response: Thank you for the good suggestion. In the newly revised manuscript, we have added the in vivo detoxification assay to confirm that the nanoreactors injection can improve the survival rate of the toxin-challenged mice. For detailed experimental methods, please see: P 20, Lines 12-16; the results have been provided in the newly revised manuscript (Please see: Figure 6f ) and shown in Figure 9. It can be seen that the nano-toxin vaccinations bestow strong protective immunity. At the same time, we provide the hemolysis induced by antibody generated from injection of PBS, RFP-CaO2@PCM@Lec, toxin, heated toxin, and nano-toxin. For the detailed experimental methods (Please see: P 20, Lines 6-11); the results have been provided in the newly revised manuscript (Please see: Figure 6d and 6e) and shown in Figure 10. Our results showed that nano-toxin has the ability to induce stronger antibodies to neutralize the toxin.  In our study, the main reason for choosing the skin punch model is that we need to continuously monitor ex vivo bacterial burden using colony counts and repair of the wound under different treatment times (For the detail methods, please refer to P 18, Lines 11-22). The MRSA was injected via subcutaneous injection, and the main purpose of skin punch is to create a wound to facilitate bacterial and bacterial toxin enrichment. Figure 5b. The flow cytometric data should be provided."

Comments-10: "The gating strategy is not described at all for the identification of germinal center B cell result in
Response: Thank you for the good suggestion. In the newly revised manuscript, we have provided the flow cytometric data in Figure 11(Please see: Figure S14 in the newly revised manuscript).

Comments-11: "Supplementary figures S5, S6, and S7 are not described at all in the text."
Response: We feel very sorry for the mistakes in the old version of the manuscript. In the revised manuscript, we have added the discussion about the S5 (please see: P 9, Lines 7-9), S6 (please see: P12, Lines 8-11), S7 (please see: P 3, Lines 6-10, in the Supporting information).

Reviewer #2
Thanks for the positive comment and constructive suggestions on how to improve our manuscript.

Comments-2: "In Figure 1e, the different absorption intensity of RFP in the groups of Free RFP (black), RFP in ethanol (red), in toxin (blue) should be explained and the corresponding content of RFP should be provided."
Response: Thank you for the good comment and suggestion. In Figure 1e, the UV absorption spectra was only used to qualitatively evaluate the RFP, which showed that the RFP were successfully loaded with nanoreactors, but the ICP assay was used to quantitative analysis of RFP and CaO2 loading rate(Please see: Table S2).

Comments-3: "In Figure 2j, why did the RFP@PCM@Lec+toxin cannot release RFP at 37 ℃? Is the size of the RFP smaller than the pore induced by toxin?"
Response: In the newly revised manuscript, ORCA program (WIREs. Comput. Mol. Sci., 2012, 2, 73) was employed to calculate the structure of RFP at the level of 6-311G (d, p), and the calculated data showed that the RFP has a diameter of 17.96 Å (please refer to the following Figure 1). The previous experimental and theoretical work indicates that most atomic long-ranged interactions are greater than 5 Å (Phys. Rev. Lett., 2004, 92, 246401); however, the α-toxin pores are estimated to be 1-2.5 nm in diameter (Biol. Cell, 2006, 98, 667), thus RFP appears overly large to pass through even the largest pore.
At the same time, due to the lack of CaO2 in the RFP@PCM@Lec nanoformulations , there is a lack of gas in the system as a driving force to promote the release of RFP, which further confirms the role of gas drive in the controlled release of drugs. Figure 12. The structure of RFP at the level of 6-311G by using ORCA program. The distance between the two atoms at the edge of the blue wire frame is 17.96 Å. The red ball represents oxygen atom, white ball represents hydrogen atom, grey ball represents carbon atoms, and blue ball represents nitrogen atom.

"How about the structure of nanomachines at 37 ℃ for 150 min?"
Response: In the newly revised manuscript, we have supplemented relevant experiments and provided the representative images demonstrating that the change trend of nanoreactors (RFP-CaO 2@PCM@Lec) structure with the prolongation of toxin action time. The results in Figure 13 showed that the long-term treatment (24 h) leads to complete collapse of the nanocomposite structure; meanwhile, we have used the immunogold staining to confirm the presence of toxin protein.

Comments-4: "The clarity of the pictures in the article is not enough, the resolution of the pictures should be improved. Some pictures are wrong, for example, the plates in figure 3a are arranged differently (In Figure 3a, some images of group I and group II are placed repeatedly, please check out and correct it), and in figure 4, the descriptions of b and c are opposite."
Response: We feel sorry to make this mistake in the old version. In this revision, we have checked the manuscript carefully, including each picture, and all the modified portions are highlighted in yellow.

Comments-5: "The format of the references should be checked."
Response: Many thanks for the good suggestions. In this revision, we have checked the manuscript carefully, including each reference, and all the modified portions are highlighted in yellow.

Comments-6: "If possible, repeatability measurements (error bars) should be included in all episodes, and they need to put their overall impact results in the background. For example, there has been a significant reduction in the number of MRSAs after treatment, but they do not provide a reference to compare their results with current studies of current clinical MRSA treatments or new antibacterial treatments for MRSA."
Response: Many thanks for the good suggestions. In the newly revised manuscript, we have added the significant difference analysis in all episodes (please see: Figure  2g, 3c, 3e, 4d, 6b, 6c, 6e). All the modified portions are highlighted in yellow. At the same time, we have provided some reference to compare our results with current studies about the clinical MRSA treatments or new antibacterial treatments for MRSA (please refer: Adv. Funct. Mater., 2018, 28, 1800011; Chem. Soc. Rev., 2019, 48, 415), ours results showed that the toxin stimulus-powered antibiotic release from nanoreactors have better antibacterial activity .

Comments-7: "How to prove that the calcium peroxide nanoparticles are coated inside the lecithin particles, rather than having some adsorption on their surface."
Response: Thank you for the good suggestions. As we all know, the CaO2 could react with water, leading to the production of calcium hydroxide [Ca(OH)2] and hydrogen peroxide (H2O2). We detected the content of H2O2 in the solution using the Hydrogen Peroxide Assay Kit. The results showed that (please refer to the Figure 2g in the newly revised manuscript), when the RFP-CaO2@PCM@Lec was incubated with the toxin at 37°C, the yield of H2O2 in solution gradually increased within the 120 min time point, and the concentration of H2O2 reached a maximum of 2.09 mmol/L, accounting for 79.15% of the theoretical production. However, when the RFP-CaO2@PCM@Lec was incubated with DI water at 37°C, the maximum concentration of H2O2 was only 0.32 mmol/L at the 60 min time point, which is only 12.10% of the theoretical production. These results confirm that most of the CaO2 was coated inside the lecithin particles.

Comments-8: "Obviously you should cite some of the latest antibacterial studies, such as the study of photothermal and photodynamic synergies."
Response: Thanks for the recommendation of these relevant papers, which are important in the antibacterial research field. These corresponding studies are cited in the manuscript as the new Refs 5 (please refer to P 3, Lines 5).

Reviewer #3
Thanks for the positive comment and constructive suggestions on how to improve our manuscript.

Response:
We are very sorry about the high similarity with other articles. In the newly revised manuscript, we have rewritten the results and discussion sections about the evaluation of immune effect of nanoreactors (Please see: P12, Lines 21-23; P 13, Lines 1-21); meanwhile, the caption of figure 5 was also rewritten (Please see: P 27, Lines 2-13). In future articles, we must strictly abide by the academic norms and avoid the recurrence of similar problems.

Comments-2: "In the manuscript line 91, nanoparticles size is 150-200 nm with relatively uniform size. However, SEM image in figure s1and TEM image in figure 1b are quite poor quality and seems to be inconsistence. Particle size and polydispersity index (PDI) is commonly measured by dynamic light scattering (DLS). Size and PDI characterization by DLS should be provided."
Response: Thank you for the good suggestion. In the newly revised manuscript, we provided the high quality SEM and TEM image (please see: Figure 1d and S2). At the same time, we have provided the size and PDI characterization by DLS in our revised manuscript (please see: Figure 2i).

Comments-3: "What is drug loading and encapsulation efficiency of RFP and CaO2 within RFP-CaO2@PCM@Lec before and after 0.22 micron filtration?"
Response: Many thanks for the good suggestions. In the newly revised manuscript, we have provided the loading and encapsulation efficiency of RFP and CaO2 within RFP-CaO2@PCM@Lec before and after 0.22 micron filtration (please refer to the Table 1). Response: Thank you for the good suggestions. In the newly revised manuscript, we have provided the antibacterial activity about different concentrations of RFP-CaO2@PCM@Lec nanoparticles, using free RFP as positive control, and the results shown in Figure 14 (please see: Figure S8) suggest that the pure RFP and H2O2 have limited antibacterial activity compared with the RFP-CaO2@PCM@Lec nanoparticles.

Comments-5: "Why did you choose free fatty acid (lauric acid and stearic acid) at the ratio of 4:1?"
Response: Thank you for the good suggestions. In our study, we have optimized the lauric acid to stearic acid ratio based on the melting temperature. The results have been provided in the newly revised manuscript (Please see: Table S1) and shown in Table 2. As shown, when the ratio of lauric acid to stearic acid is 4:1, the eutectic mixture formed with a well-defined melting point at 35.2-38.3 can meet our experimental requirements, which is consistent with the results reported in previous literature (Sol. Energy, 2002, 72, 493). So in our study, we chose the ratio of lauric acid to stearic acid at 4:1.

Comments-6: "Why did the author select rifampicin as antibiotic to treat MRSA infection? Please discuss why vancomycin which is the standard treatment for MRSA infection was not chosen."
Response: Thank you for the good suggestions. In our study, the reasons why we chose rifampicin but not vancomycin as the antibacterial agent are as follows: (1) RFP was primarily a frontline drug for the treatment of bacterial infection and showed good antibacterial activity in treating MRSA-related infections (Proc. Am. Thorac. Soc., 2004, 1, 338; Lancet, 2001, 357, 40 ); (2) In our previous research, RFP was loaded into the metal organic framework, and the combination of UV-light (365 nm), pH-triggered precise RFP release and zinc ions enables the light-activated nanocomposite to significantly inhibit MRSA-induced wound infection and accelerate wound healing (Adv. Funct. Mater., 2018, 28, 1800011); (3) We obtained the size of Vancomycin and Rifampin molecules by molecular dynamic simulation. In Figure 15, the calculated data showed that RFP has a diameter of 17.96 Å and Van has a diameter of 23.53 Å. The previous experimental and theoretical work indicates most of atomic long-ranged interactions are greater than 5 Å (Phys. Rev. Lett., 2004, 92, 246401), while the α-toxin pores are estimated to be 1-2.5 nm in diameter (Biol. Cell, 2006, 98, 667 ), indicating Van is more difficult to release, even via the largest pore. Figure 15. The structure of RFP (a) and Van (b) based on the dynamic molecular simulation.
(4) Furthermore, we used vancomycin as the control, and the drug release efficiency in nanoreactors was evaluated. The results shown in Figure 16 demonstrated that the nanoreactors for loading RFP has better drug release ability, which further validated the dynamic molecular simulation data. Based on the above analyses, we chose Rifampin instead of Vancomcin as our antibacterial agent in this study.

"Please show TEM images (negative staining) of untreated nanoparticle samples as negative control to compare with treated one."
Response: Thank you the good suggestion. In the newly revised manuscript, we have added the TEM images (negative staining) of untreated nanoparticle samples as negative control to compare with treated one (Please see: Figure 1g). The results are shown in Figure 19, and we can see that the untreated nanoparticle has a smooth surface and complete membrane structure; however, with the addition of toxin, the structure of the nanoparticle has changed significantly. In particular, when the treatment time is extended to 24 hours, the structure of the material has completely disintegrated (Please see Figure 18).  In this process, the concept of a nanoreactor was introduced for the design of a stimuli-responsive drug delivery and release nanosystem [8][9][10][11] . The potential applications of nanoreactors are not only involved in chemical synthesis, but also in many cross-cutting fields such as biomedicine 12-14 . In particular, the in vivo use of micro-/nanoreactors has attracted the attention of more and more researchers for therapy and diagnosis of various diseases 15, 16 . For construction of nanoreactors, the substrate and product should be exchanged between the inner and outer regions, that is, appropriate permeability is required for the wall of nanocompartments 17 . Moreover, the encapsulation of a wide variety of catalytic materials is another essential challenge.

A list of changes
Despite the development of several nanoreactor systems, problems still remain in the encapsulation process and permeation of the substrate and products 18 (Figure 1c and Table S1), then the DSPE-PEGylated lecithin (Lec) was used to coat the eutectic mixture and form a toxin-reactive nanoreactor for drug release, which was mixed at a mass ratio of 3:1 to prevent hemolysis and also maintain the ability to adsorb toxin. ( Figure S1).

Page5, Line 13 nanomachines nanoreactors
Page5, Line 14 Figure S1 Figure S2 Page5, Line 17 (Table S1) (Table S2) Page5, Line 17 The absorption peak of RFP can be detected at 473 nm when RFP-CaO2@PCM@Lec was dissolved in ethanol, but not when dispersed in deionized (DI) water, then the absorption peak can be restored again with the addition of toxins.
When RFP-CaO2@PCM@Lec (nanoreactors) are dissolved in ethanol, the absorption peak of RFP can be detected at 473 nm, but when dispersed in deionized (DI) water, the absorption peak cannot be detected, and then absorption peak can be recovered by adding toxin.  (Figure 2a). The results indicated that 100 μg of the nanomachines was sufficient to capture 4 μg of Hla, and further experimental results 100 μg of the nanoreactors was found to be able to capture 4 μg of toxin ( Figure  2a and Figure S3). The immunoglod staining experiment showed that the nanoreactors without toxin treatment did not display any specific binding, while toxin-treated nanoreactor surface could combine very distinct gold nanoparticles. These results indicate that showed the structural integrity of these toxins was not affected by the nanomachines (Figure 2b). nanoreactors can efficiently capture toxins without affecting their structural integrity. (Figure 2b). Page6, Line 13 nanomachines nanoreactors Page6, Line 14 nanomachines nanoreactors Page6, Line 14 8-aminonaphthalene-1,3,6trisulfonic acid disodium salt (ANTS) and p-xylene-bis-pyridinium bromide (DPX), which are used as a pair of fluorophore/quencher to evaluate the stability of liposomes 18, 26 .
In vitro antibacterial activity (Figure 3e-3f) tests also showed that nanoreactors have efficient antibacterial ability (3.02 Log), and similar results were observed in live/dead staining ( Figure  3g). Furthermore, we evaluated the antibacterial efficiency of RFP and CaO2 at different concentrations, and the pure RFP and H2O2 ( Figure S8 Figure 4d and 4e.

nanoformulations
(pure nanoreactors, free toxin, heat-inactivated toxin (heated toxin), and nanoreactors detaining toxin (nano-toxin)). As shown in Figure 4c and 4d, the untreated free toxin has high hemolytic efficiency, but after the toxin is captured by the nanoreactor, the hemolytic rate decreases significantly. Moreover, the toxicity of different nanoformulations was assessed using the terminal deoxynucleotidyl transferase dUTP nick end labelling (TUNEL) assay and the results are shown in Figure  4e and S10. Page10, Line21 nanomachines nanoreactors Page10, Line22 nothing The ability of the nanoreactors to neutralize α-toxin was further demonstrated in vivo by subcutaneous injection of pure nanoreactors, free toxin, heated toxin, and nano-toxin beneath the right flank skin of mice. Based on the skin lesions shown in Figure S11, the pure toxin induced demonstrable edema and inflammation with the extension of time (7 d, 14 d, 21 d), and this phenomenon became more and more serious, with obvious suppuration and muscle rot being observed in the skin tissue at the toxin injection site after 21 days of treatment.
However, the nanoreactor-toxin showed no significant damage to the skin. Furthermore, the H&E, immunocytochemistry (IHC) and blood routine assays were used to evaluate the toxicity of different nanoformulations at 21 days post injection. The toxin treatment was shown to induce stronger tissue damage, inflammation or lesion by H&E and IHC analysis (Figure 4f), in contrast to a similar result between the nanoreactor-toxin and the control, which was further supported by the analysis results of blood routine ( Figure S12). All the above test results reveal that the nanoreactors can effectively neutralize toxins without causing significant cytotoxicity or physiological toxicity.

Page11, Line13 nanomachines nanoreactors
Page13, Line19 RFP-CaO2@PCM@Lec nanoreactors Page13, Line19 Figure 6a   Following the safety assessment, we studied the ability of the nanomachines to elicit potent humoral immunity (Figure 5a). The induction of germinal centers in lymph nodes is one of the key steps in the immune response to infection, and affinity-based maturation of B cells occurs in these regions 36 . Our study showed that the nanotoxoid

Immune effect and in vivo
In the present study, the nano-toxin Page13, Line14 compared to than Page13, Line15 nothing Furthermore, the in vivo toxin neutralization ability of nanoreactors was evaluated by measuring hemolysis ratio (Figure 6d and 6e). It can be seen that the nanoreactors have better toxin-neutralizing ability and can significantly reduce the hemolysis rate. Finally, the protective immunity bestowed by the nanoreactors was evaluated by subjecting the vaccinated mice to toxin administration at a toxin dose of 120 μg/kg 45 , which resulted in 100% mortality within 2 h in the unvaccinated group. Meanwhile, the nano-toxin boosters improved the survival rate to 100% versus an 80% survival rate for the heat-treated toxin vaccine with boosters (n=10) ( Figure  6f) The PCM solution (3 mL) was mixed with the desired payloads (500 μL 12.5 mM of ANTS and 500 μL 45 mM of DPX in DMSO)21, 26 and then added dropwise into the preheated phospholipid solution, followed by vigorous vortex for 2 min. After cooling in ice water for 60 min, the cloudy solution was centrifuged for removing the un-encapsulated molecules and then filtered through a 0.22 micron filter. After washing three times with water, the resultant nanoreactors were suspended in water at 4°C for further use.

Page16, Line3
RFP-CaO2@PCM@Lec was nanoreactors were Page16, Line6 nothing Evaluation of toxin adsorption and hemolysis of nonareactors by using lecithin and DSPE-PEG nanomaterials at different ratios. BCA Protein Assay Kit was used for quantitative detection of the adsorption of toxins by materials. Briefly, 200 μL of 500 μg/mL nonareactors synthesized in different mass proportions ( Lec : DSPE-PEG=1:1,3:1,6:1,9:1, 12:1 and 1:0) was mixed with 10 μL of 400 ug/mL toxin to interact with each other at 37℃ for two hours, using PBS as a control. The mass of the adsorbed toxin was calculated by the absorbance at 462 nm according to the detection method of the BCA kit. Under the same experimental protocol, the hemolysis rate of the material can also be calculated by the following formula. Briefly, 150 μL of different materials synthesized at different proportions (Lec: DSPE-PEG=1:1,3:1,6:1,9:1, 12:1 and 1:0) and 150 μL of 2% RBCs were incubated for 30 min at room temperature.
After centrifugation at 2 000 x g for 5 min, the hemolysis was determined for each sample by measuring the absorbance of the supernatant at 540 nm using a microplate reader. A 100% lysis control was prepared by treating RBCs with Triton X-100. The hemolysis rate of each group was calculated as follows.
Hemolysis rate = Abs(experiment) Abs(X − 100) × 100% (1) Page16, Line21 nothing Bacterial culture. Briefly, 200 μL of 108 CFU/mL bacteria was incubated with different concentrations of nanoreactors, RFP and CaO2 at 37°C for 2 h at 120 rpm. To evaluate the bacterial mortality, the treated bacteria were diluted and uniformly plated in Luria-Bertani (LB) solid medium, followed by incubation at 37 °C for 24 h. Finally, colony forming unit (CFU) was counted and compared with the control plate. Each treatment was prepared in triplicate and the mean values were compared with each other.

Page17, Line4
nothing In Vivo Safety. Briefly, the BALB/c mice (6-8 weeks old) were first shaved to remove the hair on the back. Subsequently, 200 μL of 100 μg/mL of nanoreactors (20 μg) was injected subcutaneously, using PBS as a control. At 24 h post injection, the mice were euthanized, and the internal organs (heart, liver, spleen, lung, kidney) were collected for histological analysis by hematoxylin and eosin (H&E) staining. Meanwhile, the plasma was collected for biochemical indicator detection (ALB, ALP, ALT, AST, A/G, BUN, GLOB, TP).
Assessment was also performed on the toxicity of nanoreactors (100 μg), toxin (4 μg), heated toxin (4 μg,70℃ inactivated for 1h) and nano-toxin (4 μg toxin absorbed by 100 μg RFP-CaO2@PCM@Lec) using PBS as control. Briefly, BALB/c mice were first shaved to remove the hair on their back and the above materials were injected subcutaneously and separately to each group of mice. At 24 h post injection, the mice were euthanized, and skin samples at the injection site were collected for histological analysis by hematoxylin and eosin (H&E) and TUNEL. TUNEL staining and Ipwin32 software were used to count the number of cells with a different color fluorescence.
After 21 days of immunization, hematoxylin and eosin (H&E) skin staining and immunohistochemistry (IHC) were performed on the dorsal skin of each group to judge the viable toxicity of different treatments. At the same time, the blood of the mice was collected, and blood routine tests were performed to observe the number of white blood cells (WBC) and neutrophils (Gran).

Page18, Line1
nothing Inhibitory effect of nanoformulations on hemolysis. The ability of nanoformulations to prevent hemolysis was investigated under five different experimental groups: PBS, nanoreactors (100 μg), toxin (4 μg), heated toxin (4 μg ,70℃ inactivated for 1h) and nano-toxin (4 μg toxin absorbed by 100 μg nanoreactors). Briefly, 150 μL of different materials and 150 μL of 2% red blood cells (RBCs) were incubated for 30 min at room temperature, followed by centrifugation at 2 000 x g for 5 min. Next, the hemolysis of each group was determined by measuring the absorbance of the supernatant at 540 nm using a microplate reader. Meanwhile, a 100% lysis control was prepared by treating RBC with Triton X-100.
Finally, the hemolysis rate of each group was calculated according to formula 1.
The authors have vastly improved the content and scholarly presentation in the revised manuscript, and the data and narrative in the present version is of sufficient quality for publication in Nature Communications. Some minor suggestions remain, however, and should be addressed.
1. The authors mention the use of "DSPE-PEGylated lecithin" to prepare the nanoreactor. The term is confusing and I believe the authors mean DSPE-PEG and lecithin as two separate entities. This should be corrected. figure 4d and figure 6d do not match. In figure 4 d, the nanoreactor/toxin complex is shown to have residual hemolytic activity but in figure 6d, no hemolysis was observed. The inconsistency should be corrected.

Data between
3. Figure 6b has a high overlap of error bars, which makes the reported statistical significance hard to believe. The authors should double check the statistical analysis and data presentation.
Reviewer #2 (Remarks to the Author): Authors have revised the paper carefully following the comments of the reviewers point by point.
It can be published as is.

Reviewer #1
Thanks for the positive comment and constructive suggestions on how to improve our manuscript.

DSPE-PEG and lecithin as two separate entities. This should be corrected."
Response: Thank you for the good suggestion. In our study, Lecithin and DSPE-PEG3400 were used to coat the eutectic mixture of two fatty acids as a gate material in fabricating toxin-responsive nanoreactors for drug release. DSPE-PEG and lecithin as two separate entities. In the newly revised manuscripts, we have corrected the "DSPE- In the figure 4d, we have evaluated the hemolysis induced by PBS, RFP-CaO2@PCM@Lec nanoreactors, toxin, heated toxin and nanoreactor/toxin (the detailed experimental methods, please see: P16, L6-20), it can be seen that the nanoreactor/toxin have lower hemolysis ratio compare with the pure toxin when these materials were directly incubated with red blood cells (RBCs), these results revealed that the toxin were effectively captured by the nanoreactors so that significantly inhibiting hemolysis. However, in the figure 6d, it display the hemolysis ratio assay of antibody induced by different materials (the detailed experimental methods, Please see: P 20, L 8-13), it can be seen that when the toxin were captured by nanoreactors which have better toxin-neutralizing ability and can significantly reduce the hemolysis rate, thereby the figure 6d the nanoreactor/toxin complex have no hemolysis was observed.