Serum apolipoprotein A-I potentiates the therapeutic efficacy of lysocin E against Staphylococcus aureus

Lysocin E is a lipopeptide with antibiotic activity against methicillin-resistant Staphylococcus aureus. For unclear reasons, the antibacterial activity of lysocin E in a mouse systemic infection model is higher than expected from in vitro results, and the in vitro activity is enhanced by addition of bovine serum. Here, we confirm that serum from various species, including humans, increases lysocin E antimicrobial activity, and identify apolipoprotein A-I (ApoA-I) as an enhancing factor. ApoA-I increases the antibacterial activity of lysocin E when added in vitro, and the antibiotic displays reduced activity in ApoA-I gene knockout mice. Binding of ApoA-I to lysocin E is enhanced by lipid II, a cell-wall synthesis precursor found in the bacterial membrane. Thus, the antimicrobial activity of lysocin E is potentiated through interactions with host serum proteins and microbial components.

the authors, the lipid binding domain was removed. First, most of apoA-I is able to bind to lipids, so the terminology of lipid binding domain is ambiguous. Second, it would help if the authors clearly describe with amino acids have been removed in this truncation variant (in the result section). Based on the description in the Methods, it seems that the first 65 residues and residues 146 to 160 were removed from apoA-I (I could be off by a couple of amino acids). What was the rationale for removing these specific residues? The C-terminal domain (residues ~ 190 to 243) is considered the part of the protein that initiates lipid binding, aided by its relatively unstructured and make direct contact with lipids.
A related point is that the C-terminal part of the protein has been identified as an important site 243 (Apolipoprotein A-I exerts bactericidal activity against Yersinia enterocolitica Serotype O:3*, 2011, J. Biol. Chem 268, 38211). Similar to the present study, the antimicrobial activity of apoA-I was not direct but needed the complement system. The discussion and wider scope of their findings will be strengthened if the authors include a brief discussion on this.

Minor points
Since the major focus of the study is apoA-I, the authors may consider including this into the title.
Line 86, purified human apoA-II was used, it is not clear if this also recombinant protein or isolated from plasma (and how is was isolated).
In the Discussion, line 258, it is mentioned that lipophorin has similar functions as apoA-I. However, the lipophorin studied was apolipophorin I and II, which are apoB like apolipoproteins in insects with masses of 220 and 74 kDa, and not to be confused with the much smaller apolipophorin III (~18 kDa). Apolipophorin III shares similar structural and functional characteristics compared with apoA-I. The manuscript "Serum proteins potentiate therapeutic effect of lysocin E against S. aureus " by Hamamoto et al. reports on the impact of Apolipoprotein A1 to increase the antimicrobial activity of the antibiotic lysocin E, previously reported to interact with menaquinone and/or lipid II. The authors claim to provide first evidence that antimicrobial activity can be potentiated through interactions of host derived factors, i.e. serum protein ApoA1, and microbial components, i.e. lipid II. At first glance the study appears interesting. However, the data presented do not support the proposed model. While the manuscript is well written and the (individual) results are properly described the model derived is superficial and flawed. Importantly the authors appear to ignore current knowledge on the multifactorial roles of apolipoproteins. Although the main role of HDL and its principal apolipoproteins has long been considered to be its participation in reverse cholesterol transport and its anti-atherogenic effect, more recent studies have involved these proteins in other defensive functions in mammals, such as antiviral, antimicrobial and antiinflammatory activities. Apolipoproteins are crucial factors of the innate defense and share characteristic features with cationic antimicrobial peptides and their abilities to bind to bacterial membranes and to adopt specific secondary structures in membrane environments, an essential prerequisite to their attachment and insertion into bacterial membranes. The role of menaquinone is still uncertain and it is unclear if lipid II binding to ApoA-I, given it is specific, is the reason for the potentiating effect. As shown for e.g. albumin, the binding capacity is strongly influenced by fatty acid and other factors. Therefore, other lipids such as phosphatidylglycerol might have the same effect and are basically co-factors for ApoA-I in binding lysocin-E. The interaction of apolipoproteins with (phospho)lipids has been demonstrated (e.g. lipoteichoic acid, sphingomyelin, LPS, cardiolipin, phoshatidylglycerol). I deem it necessary to provide further experimental data on the overall binding capacity of ApoA-I for lysocin-E, lipid II and most importantly other lipids mentioned from the authors themselves. Otherwise the model is highly speculative and not sufficiently supported by the experimental data. In terms of consistency, the authors should also stick with one species as protein binding is species dependent. Therefore, it would have been better to use heterlogously produced mouse ApoA-I and -II for the in vitro experiments. The authors use human recombinant ApoA-I and -II ( fig. 1b), isolation was done from Bos taurus (ext. data fig. 2) and in vivo experiments were done in mice. It is very difficult to pool these results and extract the correct information.

Specific comments:
Page 2, 37-41: The authors claim to demonstrate the potentiating effect of serum components on antimicrobial activity for the first time. The potentiating effect of serum on antimicrobial activity was already shown for 3. generation cephalosporins in 1988. https://pubmed.ncbi.nlm.nih.gov/2496656/ Although they aim to connect novelty to an interaction with an additional microbial component, namely lipid II, does not justify this statement. Firstly, the authors could not provide conclusive data on a specific interaction with lipid II. Second, a direct correlation between the observed therapeutic effect and the proposed players involved (i.e. ApoA1, Lysocin E, lipid II) is not proven. Page 2, 46-48: The authors claim that protein binding (PB) is generally considered a problem in terms of antimicrobial efficacy. The effect of PB on compound activity is controversially discussed and still matter of ongoing research. Regarding antibiotics, it was shown that Daptomycin has a PB of over 90%, but its clinical efficacy is not impaired as predicted from the in vitro data, because they are administered at higher doses than necessary. The same is true for telavancin where a high PB has no effect on clinical efficacy. Furthermore, PB can also act as a depot for certain substances, which in the case for antibiotics, can keep the concentration above the MIC for a longer period of time. Authors should take care in using words such as "generally" if there is ongoing research on the topic.
Page 2, 53-59,   The impact of serum on the antimicrobial activity of daptomycin and vancomycin was tested. While Ext Data Table 2 includes nisin and vancomycin, daptomycin was not included. Daptomycin should also be tested in presence of ApoA1. What is the rationale for choosing 25 µg/ml ApoA1? Why did the authors use human plasma instead of human serum when comparing the effects on antimicrobial activity? While there is no change in MIC comparing plasma with serum values are questionable. Figure 1b,c: The panel shows that ApoA-I and ApoA-II both potentiate the effect of lysocin E in vitro, yet in vivo the effect was attributed to ApoA-I. While this might be a concentration dependent effect (7fold less ApoA-II), I would also include ApoA2 gene knock-out mice to rule out ApoA-II is necessary for the ApoA-I effect on lysocin E activity. Do the authors have any explanation for the lack of in vivo activity of ApoA-II? S. aureus Newman was used in the mouse model. Include MIC in Ext Data Table 1. Recommend to include a control where the authors show the effect of serum in the absence of lysocin E Ext. Figure 2: There are additional bands visible between 36 and 64 kDa. It is recommended to identify these proteins as well.
Page 10, line 141: It is not surprising that ApoA1 binds to S. aureus, as ApoA1 interacts with membranes and was shown to interact with LTA. Only relative numbers are given. How much ApoA1 did bind to cells? The authors should further provide raw data.
Ext. Data Fig 4: The authors used BLI for determination of binding parameters. Please provide evidence that the BLITZ system is suitable for the determination of binding parameters of small molecules, such as lipid II. Nisin could be used as a control and binding parameters have been determined by different methods. Binding parameters of lysocin E to lipid II indicate no high affinity interaction, compared to other lipid II binding antibiotics, e.g. Nisin Kd 2.68x10-7 M (DOPC + Lipid II) ; Nisin Kd 1.03x10-6 M (DOPC only).
Ext. Data Fig 4d: How was the pull down assay performed? Controls should be included.
Ext. Data Fig. 5: In lines 266-268 the authors correctly mention, that ApoA-I binds to phosphatidylglycerol (PG) and cardiolipin (CL), therefore, I do not understand why they did not use vesicles made of PG and included CL. This would be an easy experiment and would provide data on the necessity and specificity of lipid II.
Page 17, 278 ff: The authors fail to provide evidence for the role of menaquinone in their model. As shown by Santiago et al. the binding sites of MK and lipid II overlap and the high level of resistance towards lysocine E in menA and menB mutants is attributed to the slow growth of these strains. A slow growth also means that less lipid II is present and therefore, the effect of lysocin E is diminished. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5964011/pdf/nihms946863.pdf Such effects have further been reported for S. aureus small colony variants.
It is more likely that the lipid tail is responsible for ApoA-I intercation. While both, MK and lipid II, contain a lipid tail, the one of lipid II might bind more strongly to ApoA-I making MK the less favorable interaction partner. For the latter it would be also crucial to show which lipid II species (including pentaglycine and amidation ?) was used? Vancomycin shows different binding affinities to modified vs unmodified lipid II. The authors should further include controls, e.g. cardiolipin or phospatidylglycerol. Vancomycin binds to the D-ala-ala terminus of lipid II and does not insert into the membrane (compared to nisin). Vancomycin is not the ideal compound for comparison.
Lipid II: Please provide evidence on the identity and purity of lipid II by mass spectrometry. It is further recommended to use more direct methods to determine the concentration of lipid II. The authors incorporated lipid II into liposomes. It should be determined how much lipid II has been incorporated. This manuscript by Hamamoto et al provides a mechanistic understanding of how the in vivo efficacy of the novel Lysocin E antibiotic is greater than would be expected on the basis of its in vitro activity towards methicillin resistant Staphylococcus aureus (MRSA). Although interaction with host proteins commonly reduces the in vivo efficacy of antibiotics, the minimum effective dose of Lysocin E, a lipopeptide antibiotic, was potentiated through its interaction with the serum protein ApoA-1. The in vivo relevance was further demonstrated using ApoA-1 deficient transgenic mice, in which the enhanced in vivo efficacy of Lysocin E was no longer evident. The authors further reveal that the interaction of lysocin E with ApoA-1 was enhanced by Lipid II, a lipid carrier of peptidoglycan subunits that is essential for cell wall synthesis. This adds to previous knowledge that lysocin E targets menaquinone, which is an essential component of the electron transport chain in Gram positive bacteria. The authors present a model whereby interaction of lysocin E with ApoA-1 in serum promotes greater interaction with both menaquinone (MK) and lipid II, causing maximum membrane disruption. The work is carefully done with a strong combination of advanced biochemistry techniques and infection models. There is support for the authors claim of being the first to demonstrate that antimicrobial activity can be potentiated through interactions of host serum proteins with microbial components to enhance the therapeutic effect. However, some of the claims seem to be contradictory, or stated in a manner that the meaning is unclear.
Comments and questions for the authors to consider: 1. The present work identifies lipid II as a target for Lysocin E, and the abstract claims that the binding capacity of Lysocin E to ApoA-1 was enhanced by lipid II. This is somewhat confusing, since lines 142 to 145 state that the enhancing effect of ApoA-1 cannot be explained by increased accumulation of lysocin E to the cell surface, while lines 171-172 state that "lipid II increased the binding of lysocin E to ApoA-1 in a pull down assay in the presence of menaquinone". This seems to suggest that ApoA-1 can promote increased accumulation of lysocin E. Taking this into account, could the authors please provide a temporal view of how lysocin E is bactericidal during antimicrobial therapy? For example, ApoA-1 is in serum, and lysocin E has a binding constant of 3.6 µM for ApoA-1 compared to 4 µM for menaquinone (MK). How would this relate to therapeutic levels of lysocin E in blood, and should we assume that during antimicrobial therapy, lysocin E forms a complex with ApoA-1 before it encounters a microbial surface? Once this complex engages lipid II on a microbial surface, could it not then recruit more lysocin E? 2. Would it be possible for the authors to image lysocin E on the microbial surface in the presence and absence of ApoA-1? For example, they have been successful in biotinylating lysocen E. Presumably a fluorescent streptavidin derivative could be used to quantify lysocin E binding to cell surfaces when bacteria are pre-treated with ApoA-1 compared to non-treated cells. Assessing overall fluorescent intensity, accompanied by microscopy to visualize fluorescence localization might provide some valuable mechanistic detail.
3. ApoA-1 has a lipid binding domain, and its ability to enhance the MIC of lysocin E is attenuated when this lipid binding domain is deleted. It is presumed that this is due to loss of binding to lipid II. However, lysocin E is also a lipopeptide. The authors could consider some assays to assess the this variant to bind to S. aureus cells. Is it strongly reduced? 4. The authors use Nisin as an example of another antimicrobial that interacts with lipid II and report that as with lysocin E, the activity of Nisin is also enhanced by ApoA-1. Teixobactin is a newly described antimicrobial that also targets lipid II. Could the authors comment on whether ApoA-1 would also be expected to potentiate the activity of Teixobactin? The discovery of Teixobactin was also published in Nature, and it is a significant omission that this is not mentioned.
Minor comments: 5. Binding of lysocin E to ApoA-1 was conducted in the presence of menaquinone. Should the authors have also tried menaquinone as a co-factor in binding of lysocin E to lipid II? 6. Extended data Table 1 shows MIC data for menA and menB deficient strains of S. aureus, but there is no mention of this in the text.
Reviewer #4 (Remarks to the Author): The manuscript by Hamamoto represents a very interesting and original report showing that the activity of a recently identified antibiotic (lysocin E) is strongly potentiated by a host protein.
Overall, the data are of strong significance to the field of infectious disease/microbiologists and could have an impact on the development of antibiotics. Overall the paper is well-written, although the paper would benefit from a more thorough introduction and better-balanced discussion also discussing other reports suggesting synergy between host immune components and antibiotics.
Main comments: 1. Number of strains used to show synergy between ApoA-I and lysocin E is limited. In Extended Data Table 1, only four S. aureus strains were tested (including an ATCC strain and a laboratory strain (RN4220)). This makes it difficult to understand how broadly applicable these data are for clinically relevant S. aureus strains. Did the authors just pick S. aureus strains for which this works? It would be more convincing if the authors also include clinical isolates, including both MSSA and MRSA (including the highly virulent USA300). Suggest to determine whether synergy between a) ApoA-I and lysocin E AND b) bovine serum and