Role of NS1 and TLR3 in Pathogenesis and Immunity of WNV

1.

under 2.4 Animal and   protection study: Inoculation route of mice with WNV should be mentioned.

Corrected (i.p.)

2.

Overall in the manuscript:   Check spaces between number and unit

Thank you. Corrected

3.

Overall in the manuscript:   Make sure that "µ" is used instead of "u" in Units

Corrected

4.

Line 235: Cell number is   probably 5x10⁵ not 5x105

Thank you. Corrected

5.

Line 237: "CO2"   should be written as CO₂

Corrected

6.

Line 261: "Sybr"   should be written as "SYBR"

Corrected

7.

Line 268: "DDCt"   should be written as "ΔΔCt”

Corrected

8.

Line 332: Closing bracket at   end of paragraph is missing

Corrected

9.

Fig.2: Scaling of the y-axe:   negative values are not needed in A – D

Corrected

10.

Fig. 4: Labelling of graphs   is missing (A - E)

Corrected

11.

Line 607 and 615: Format of   citations should be checked

Corrected

12.

Supplemental Material: Fig   S2: Part B is not readable. Labelling A) and B9 should be same size. Several   times in Supplemental material “ELISA antigen production”: Centrifugation   speed is written as 3000g, 3000xg, 3000x g….

Corrected

1.

Overall, this research is   very interesting. Confirms that TLR3 plays a role in brain cell infiltrates   during WNV infection.

Thank you

However, the clinical   relevance of this should be discussed in more detail.

We addressed this issue in the revised   manuscript; Lines 51- 53 and 596-602.

If NS1 vaccination   compensates for a lack of TLR3, how does this translate to a useful vaccine   for a population?

Addressed in the Discussion and in Conclusions,   particularly Line 612.

What pathways could be at   work that could be targeted in the future? What is the future direction of   this research?

Addressed on Lines 572 onwards and 587   onwards.

2.

The introduction is very   long and must be shortened. The paragraphs are repetitive in several ways and   could be more concise.

We reduced the introduction by approximately   25% according to the suggestions of the reviewer.

Lines 97-100 provide   important information, but the thought is not well developed. It is unclear   why this is necessary information at this stage of the intro.

Shortened and reworded more concisely

Lines 102-105 are data   unnecessary for the intro.

Deleted

3.

Line 150: “The virus” is   unclear which strain you are discussing

Clarified: WNV ITA09

4.

Line 152: add   city/state/country info for the cells discussed

Done

5.

Line 173: To confirm, 105   pfu was used, not 10^5 pfu?

Yes, 10^5. Thank you. Corrected.

6.

The thorough description of animal   signs of disease is appreciated, perhaps a table would make it easier for the   reader to follow.

New Table 2, replacing the original text

7.

Missing a section to   describe the statistical analyses performed and software used for each test.

Inserted in Materials and Methods as section   2.13.

8.

Figure 2: Why do the y-axis   have a negative OD value reflected?

Good point. Negative OD values deleted.

9.

Figure 3: It would be   helpful as a reader to see these survival curves combined to compare WT and   TLR3-/- mice survival. If this makes the graph too busy, try doing smaller   panels to reflect the different vaccines used.

We re-designed Figure 3, making two panels   and leaving out all groups with 100% survival (these are mentioned in the   text).

10.

Lines 420-422: a table to   represent this would be helpful for the reader

Table 2 inserted.

11.

Figure 4: The y-axis is not   explained, the color key is too small

Figure legend modified to explain y-axis.   Color key increased from font size 9 to 12.

12.

Section 3.3.3: Several typos   are throughout this section.

This is now Section 3.4.3

Additional question: what   was the justification for using plasma only? WB is known to have detectable   virus longer than other blood fluid fractions, did the authors consider evaluating   this?

We agree. However, we opted for the cell-free   variant in order to detect viral RNA from cell-free virus particles, which   most closely reflect infectious virus.

13.

Lines 455-459: What is the   significance of this? It appears that, statistically, there is no difference   here, and the section is verbose

The paragraph about viral spread was strongly   edited and supplemented with a new Table 3

14.

Lines 462-466: what organs   are you discussing here? It is unclear as the explanation progresses

In principle, the blood stream may bring WNV   to each single internal organ. For our study, we selected spleen (easy to   test but not so relevant for the disease) and brain (most relevant for WNV   disease).

15.

A table to summarize these   results would make it easier to understand. The text is difficult to follow.

New Table 3 inserted.

16.

Lines 485-489: This is not   necessary, present the results without a miniature intro.

Deleted

17.

Lines 509-512: This seems to   depend on the results of one single animal. How relevant is this?

As explained in the revised text (Lines 474   onward), we used ranking of the animals according to various parameters as a   help to understand the data. Some observations seemed to be noteworthy even   without statistical significance.

18.

The summary starting on Line   537 is good, but the wording of the entire results section can be confusing.

Paragraph reworded to clarify (Lines 502   onward)

19.

Lines 554 (and elsewhere):   appears that special characters have been removed and are blank.

Corrected

20.

Figure 9: All other figures   were clearly created using a different software. This figure is a lower   quality and fuzzy to read. In the text, correct R2 to a superscript 2.

Figure 9 revised; superscript corrected

21.

Intro paragraph needs work.   As a whole, the writing of this section needs to be more concise.

Entire discussion revised

22.

Line 615: “Neuroinvasion of   WNV was the primary cause for various inflammatory signs in the brain”. This   is a very obvious statement and is supported by years of other research. Instead,   discuss what signals were initiated by neuroinvasion.

Entire discussion revised

23.

The authors need to discuss   more about NS1’s role when TLR3 is absent. This is currently presented as an   interesting point, but the application is not discussed. Potential pathways   could be discussed.

Entire discussion revised. Relevant   references included. Potential pathways addressed

24.

How does this relate to WNV   infections in individuals that don’t have TLR3 deficits? What else do these   data tell you?

Apparently, immunization against NS1 is less   relevant in individuals with an intact TLR3 pathway, yet, gE and NS1 antigens   seemed also to adjuvant each other (Figure 2, Table 3, Clinical score,   Conclusions)

25.

Be consistent with the way   numbers are written. Generally, <10 is written in text format, while   numbers >10 are numerical in the text. This is not consistently done   throughout the manuscript. “Thus” is overused and repeated (ex: Lines 472 and   475).

Corrected

26.

The paper has interesting   and valuable data, but needs more clarity when presenting the information in   the text.

Done

1.

Low number of individuals   per group and lack of sound statistics in many of the analyses performed:   Five individuals per group may be insufficient if the aim is to detect   significant differences between treatments. This problem affects the whole   study and can hardly be solved now. I understand the ethical constrains in   numbers of animals used in experimental work, but indeed, choosing   insufficient number of individuals for an experiment can be unethical if no   clear conclusion over the experiment is to be obtained.  I do not mean   that the work is unreliable. By opposite, I believe that the main conclusions   are well supported, mostly due to the thorough experimental work done using a   wide range of methodologies intended to better characterize the different mechanisms   displayed during WNV infection.

We agree

But the authors need to bear   this gap in mind throughout the text to avoid misinterpretations that are not   formally supported statistically. For instance, survival numbers obtained in   each group are somehow indicative, but there is no statistically supported   difference between groups. Same for seropositive individuals, etc. Under this   perspective, the word “difference” needs clarification: e.g. in the results   section, paragraph 3.2.2 (“Neutralizing antibodies”), the sentence “An   interesting difference…” pointing out that 3 over 5 is different from 5 over   5, the authors should think: “how many times I would expect to get this same   difference if I repeat the experiment?”  As experimental variation in   animal models is usually high (an interesting discussion concerning   experimental variation in mice inoculated with WNV can be found in   Pérez-Ramírez, E. et al, JGV 2017; 98:662-670) I would suggest the authors to   acknowledge this limitation in the text (especially in the discussion section)   and indicate that further experimentation (with larger groups) is needed to   statistically confirm the observed results.

We addressed these issues in the discussion   (Lines 469 onward), including also the mentioned reference

2.

Anti-NS1 antibody   determinations. The data obtained using the ELISA developed by the authors to   detect anti-NS1 antibodies, which are presented in figure 2 A-B, are very   doubtful. I do not mean the ELISA is not trustful. In fact its development is   elegant and apparently correct. My concern is over the results and not over   the ELISA. One issue is that the results shown in figure S4 c) barely agree   with those in fig 2 A-B. In the first, almost all sera reach O.D. > 0.5   and the mean is well above this value, whereas in the second the OD values   are <0.5 in almost all cases. Is there any explanation for this   disagreement?

It is true, the same amount of purified NS1   antigen induced a higher antibody response in the preliminary experiment (Fig.   S4C) than in the actual immunization/challenge experiment (Fig. 2B). We   cannot really explain this difference. However, this is the reason why we   speak in our manuscript rather of “prior immunization with NS1” than of   “antibodies against NS1”. The fact that our NS1 was actually immunogenic (able   to induce antibodies) in the main experiment has been shown, though, in the   combined immunizations (NS1+gE). Antigen aging or immunogenetic differences   due to antigen batch properties would probably be poor explanations.

Indeed, the y-axes scale   should be chosen to better observe these differences. I suggest choosing OD   scales from 0 to 1.

Scaling in Fig. 2A,B, adapted accordingly.

Another issue is that the   values obtained are so low that the differences between groups are very   small, thus raising a question: do these small differences respond to true   differences in NS1-specific antibody levels? In this regard, it is worth to   remind that background noise in ELISAs may vary depending on overall serum   gammaglobulin levels, which are affected by treatments such as inoculation of   immune stimulators. How can the authors ensure that the small differences   observed among groups are due to NS1-specific antibodies, and do not respond   to different non-specific gammaglobulin levels (or other factors)?

We acknowledge in the text (paragraph 3.3.1)   that the antibody response of TLR3KO mice against NS1 did not significantly   differ from the reactions of the mock-immunized mice. In contrast, WT mice   clearly developed antibodies against NS1. PBS/adjuvant-immunized mice served   as controls for any unspecific increase of gammaglobulin levels.

Another recommendation for   the authors at this respect is to test sera from infected mice (e.g. if   obtained, those from preliminary assays to test WNV susceptibility in C57BL76   mice) on this NS1 ELISA, as well as on the DIII-gE ELISAs. These analyses   will serve to two different purposes: on the one hand, to validate the ELISAs   (a formal validation is lacking in this study) and, on the other hand, to compare   the titres (I would strongly recommend titration in parallel) obtained in   sera from infected vs immunized mice: by doing this you will have a more   accurate idea on how far (or close) to the physiological Ab levels (i.e. the   levels attained during a normal immune response to WNV infection) are the NS1   and gE Ab levels measured in the different experimental groups immunized with   NS1, gE, NS1 plus gE, and mock.

This is a very good point. Unfortunately, we   do not have enough sera left to do titrations on NS1 antigen. However, some   formal validation of the tests has been done throughout the preliminary   experiments (S3). Throughout these tests, the sera of gE-immunized mice had   been tested not only on gE antigen but also on NS1 antigen and vice versa for   NS1-immunized mice. The results were not different from the reactions of the   sera from mock-vaccinated mice. In the case of gE, the ELISA results were   also compared to titrations of neutralizing antibodies.

Note that although the group   vaccinated with Equip commercial vaccine would do the job for gE-specific   antibodies, this is not the case for NS1 Abs, which can hardly be stimulated   by a vaccine consisting of an inactivated virus such as Equip. To finish with   this point, please bear in mind that NS1 has shown to be a bad immunogen in   some animal models (Rebollo, B et al, Comp Immunol Microbiol Infect Dis.   2018; 56:30-33) so there is some basis for expecting a low Ab response to NS1   in NS1-immunized individuals.

With all respect, we do not agree. The equip   WNV vaccine predominantly constitutes of crude WNV-infected cell culture   supernatant, which apparently contains immunogenic levels of NS1.

Yet, we have taken up your suggestion, citing   Rebollo et al on Line 62 .

3.

Effect of the time   post-infection disregarded:  In the analyses performed in organs the   authors do not differentiate the results by the time point (dpi) they were   obtained. Obviously, studies in organs arise from individuals either   succumbing to the infection (i.e. during a time window of 6 to 10 dpi) or   euthanized at the end of the experiment (21 dpi). This is crucial to   interpret the results since, for instance, it is relatively rare to find   positive viral loads in organs in survivors, while these viral loads are   expected to be high in mice succumbing to the infection. This effect explains   a lot of the results obtained because surviving the infection is expected to   strongly correlate with low viral loads in organs, mild clinical outcomes,   low inflammatory responses (cell counts, citokines, etc) in the brain and so   on. For instance, in the results section 3.3.3 “Viral spread” (page 14) lines   458-460, the results are described as if the groups were homogeneous (“In the   TL3KO group, WNV-RNA was detected in 3 out of 5 mice vaccinated with NS1 and   all mock vaccinated mice”). The reader may legitimately wonder if those   WNV-RNA positive spleens are from those mice succumbing to the infection,   i.e. 3/5 from the NS1 group and 5/5 from the mock group (or otherwise).    In any case, the authors should indicate this fact in the text, and   probably mention this effect in the discussion.  The same applies for   other analyses in which the time points are important, i.e. those described   under sections 3.3.3 and 3.3.4.

Indeed, this is a point to consider. We   acknowledge this fact, firstly by describing the experimental conditions,   lastly also by discussing the limitations of our study (Lines 569 onwards).   For the interpretation of our results, we stratified according to presence or   absence of WNV in the brains, which corresponded well with the clinical   scores.

4.

In the beginning of the   introduction, it should be mentioned that WNV is pathogenic not only for   humans and horses but also for a wide range of wild bird species.

Done, Line 38

5.

.        “Originally   from Africa” (page 1 line 38): Better “First known to Africa” since the   origin of WNV remains unclear.

We have removed this sentence from the revised   introduction

6.

Id line 43: “…candidate   vaccines have been extensively tested…” : please add: “in animal models”.

Done, Lines 42-43

7.

Page 2 line 50:   non-structural proteins include NS2a, NS2b, NS4a and NS4b polypeptides.   Please include.

Deleted in order to focus on NS1 and gE

8.

Id, line 92: For NS1 there   is also evidence of negative effects on protection (see Rebollo, B et al.   Comp Immunol Microbiol Infect Dis. 2018; 56:30-33, above cited).

Mentioned and cited in the revised discussion

9.

Throughout the text, the   word “vaccinated” is often misused: the only true vaccinated individuals are   those receiving the commercial Equip vaccine, while mice receiving doses of   recombinant proteins expected to elicit an immune response should be better designated   as “immunized” rather than “vaccinated”.

Corrected

10.

Materials and Methods   section: page 4 section 2.3 “Virus strains and cells”: Vero  cells are   mentioned twice, and the full description is not included in the first   mention, but in the second. Please correct.

Corrected

11.

Throughout the text the   virus dose used in the inoculations reads “105” (one hundred and five)   whereas it should read 10⁵ (ten to the five, or one hundred   thousands) (similar mistake with H2SO4 instead H₂SO₄ in   page 5). Please correct.

Corrected

12.

Inoculum: at several   instances in the text it is stated that the inoculum injected to mice (10⁵   pfu/mouse) is 100x LD₅₀. However, the LD₅₀ calculated   for this strain in a previous work (cited in the text: reference #41) was   2.69 TCID₅₀. This value is consistent with that found in another   study (Pérez-Ramirez, E. et al, JGV 2017; 98:662-670, above cited) for the   closely related strain IT 15803/2008 (LD₅₀= 3.16 pfu). This means   that 100 x LD₅₀ is around 3 x10² and not 10⁵   pfu. The authors should explain this discrepancy (perhaps linked to the mouse   strain used in the experiment?). Please correct the numbers accordingly if   needed.

In our experience, LD₅₀ values may   very considerably with different virus stocks. Therefore, we determined in a   preliminary study the LD₅₀ of our stocks in the context of our   mice and our mode of inoculating (Results documented in supplementary data   S2)

13.

Note that in page 8 the   first paragraph (“Since TLR3KO mice…”) has no section heading. It should have   been numbered section 3.2, and the following sections renumbered accordingly.

Thank you. Done.

14.

Figure 4: Panel marks (A, B,   C) are lacking in the figure. Also, in the figure legend, it is unclear   whether the clinical scores are cumulative, since it reads: “Each bar   represents the clinical score of one individual mouse at its day of   euthanasia”.  Please correct to state clearly the scores are cumulative.

Figure and Legend corrected accordingly

15.

Section 3.3.4 (Page 16):   Flow cytometry results are presented in a most confusing way:  Lines   509-512 provide a good example. In addition, in Fig 6 label colours   designated groups that are difficult to identify, corresponding to   combinations of groups (gE- = NS1+ mock?) that are not easy to follow. Also   in this section, there are reiterative paragraphs that are redundant with   those in the Materials and Methods section (e.g. lines 500-505). It is   unclear the real outcome of this analysis. Please re-write. Remind the above   “major point 1”: with this low n º of individuals per group, the statistical   significance of these results by group is null. Basically, this analysis   shows that leukocyte cell counts of any type  in WNV-RNA + brains is   higher than in RNA – brains, reflecting that WNV loads in brains correlate   with higher neuro- inflammation, mortality and severe clinical outcomes,   which is expected and already shown in the other results obtained.

The complete set of data, from each single   mouse has been added in the form of a supplementary Excel Table S1 (Supplementary data 4).   Reiterations removed.

16.

Page 20, lines 554-555:   Please specify the type of IFN and TNF (Gamma? Alpha?).

Done

17.

Figure 9: It does not   specify whether the results shown correspond to WT mice, TLR3KO mice or both.   Please specify.

Specified in the Legend to Figure 9

Also, in Fig 9 B) lymphocyte   counts are depicted, but why did the authors choose lymphocytes only? Why not   monocytes too?

The various inflammatory cell populations   were similarly independent of the viral titers. We show lymphocytes as a   representative example (specified in the revised text)

18.

Discussion (page 22, lines   596-597): According to the text and Figure 2 A-B, mice immunized with gE   alone were not tested for antibodies to NS1. Were they?

Indeed, the non-vaccinating antigen was   consistently used as a control for the vaccinating antigen. None of the   animals immunized with a single antigen developed reactivity (antibodies)   against the non-vaccinating antigen. Mentioned in the revised Materials and   Methods.

19.

Id, lines 598-601. Overall,   the analysis of NS1 antibodies needs some analyses to gain reliability, as   suggested in “major point 2” (also applies for page 23, lines 644-5).

Cross-reaction and unspecific reaction were   at a negligible low level.

20.

Id line 610: As remarked   above (major point 2 and minor point e) there is recent evidence on the   deleterious effect of NS1 immunization on protection against WNV infection in   susceptible animals (birds). Therefore, the role of NS1 immunization on   protection against WNV infection remains uncertain, which should be reflected   in the text.

We did not recognize any adverse effects upon   immunizing with NS1 (inserted and referenced in the revised Discussion).

21.

Id, line 611 and 633: viral   loads instead viral titers.

Corrected systematically

22.

Id, line 634: Figures 5E and   F instead  C and D.

Thank you. Corrected.

23.

Page 23, lines 649-653:   Confusing sentences. Please , re-write and explain better. Take more space if   needed. It is the most important finding of the work and needs more careful   description.

Thank you. We revised the entire Discussion,   taking this suggestion very seriously. Paragraph re-written accordingly.

24.

.        Id, lines   658-660: The effect is clearly shown, but the mechanism behind it is neither   explained nor speculated about. Could the authors speculate on the possible   causes of this effect? Instead, may the authors provide some clues on which,   in their opinion, could be the next steps to investigate the causes behind   this effect?

Included in the revised Discussion.

25.

Conclusions: the final   sentence is very weak: the role of NS1 on WNV infection and protection is   still poorly understood, so that the assertion that NS1 should be included   into WNV vaccines could be wrong and risky, and in any case, more research is   needed to ensure this point. Indeed,  isn’t it rather odd to think on   designing vaccines to protect  TLR3-deficient people from WNV infection,   bearing in mind that no vaccine is currently in use for ordinary people? (I   ignore if this deficiency exists and, if so, which is the frequency in the   population, but I guess it’s rare).

We intended to EXTEND possible protection   against WNV to people with defects in the innate immune system, rather than   to invent a vaccine specifically for those.

However, we strongly advocate testing of such   vaccines in non-human primates prior to use in humans.

We included both of these thoughts into the   revised discussion.