Identification of the HSPB4/TLR2/NF-κB axis in macrophage as a therapeutic target for sterile inflammation of the cornea

Sterile inflammation underlies many diseases of the cornea including serious chemical burns and the common dry eye syndrome. In search for therapeutic targets for corneal inflammation, we defined the kinetics of neutrophil infiltration in a model of sterile injury to the cornea and identified molecular and cellular mechanisms triggering inflammatory responses. Neutrophil infiltration occurred in two phases: a small initial phase (Phase I) that began within 15 min after injury, and a larger second phase (Phase II) that peaked at 24–48 h. Temporal analysis suggested that the neuropeptide secretoneurin initiated Phase I without involvement of resident macrophages. Phase II was initiated by the small heat shock protein HSPB4 that was released from injured keratocytes and that activated resident macrophages via the TLR2/NF-κB pathway. The Phase II inflammation was responsible for vision-threatening opacity and was markedly suppressed by different means of inhibition of the HSPB4/TLR2/NF-κB axis: in mice lacking HSPB4 or TLR2, by antibodies to HSPB4 or by TNF-α stimulated gene/protein 6 that CD44-dependently inhibits the TLR2/NF-κB pathway. Therefore, our data identified the HSPB4/TLR2/NF-κB axis in macrophages as an effective target for therapy of corneal inflammation.

Thank you for the submission of your manuscript Identification of the HSPB4/TLR2/NF-κB axis in macrophage as a therapeutic target for sterile inflammation of the cornea" to EMBO Molecular Medicine and please accept my sincere apologies for the delayed reply. We have now heard back from two of the three referees whom we asked to evaluate your manuscript and, since the review process has been delayed, have decided to withdraw referee #2. You will see that the referees find the topic of your manuscript potentially interesting. However, they also raise significant concerns on the study, which should be addressed in a major revision of the manuscript.
In particular, reviewer #1 highlights that, in addition to other experimental concerns, the injection experiments should be controlled by the use of another recombinant protein to determine whether the injection itself causes inflammation. Importantly, reviewer #3 highlights that loss of function experiments should be included.
Given the balance of these evaluations, we feel that we can consider a revision of your manuscript if you can convincingly address the issues that have been raised within the time constraints outlined below.
Revised manuscripts should be submitted within three months of a request for revision. They will otherwise be treated as new submissions, unless arranged otherwise with the editor.
I look forward to seeing a revised form of your manuscript as soon as possible.
Yours sincerely, Editor EMBO Molecular Medicine ** Please register for the EMBO Molecular Medicine Conference: Molecular Insights for Innovative Therapies (1st -3rd December 2011) by November 15th. As a valued author you or members of your laboratory are eligible for a special discount price available only at the following link: http://www.embl.de/training/events/2011/EMM11-01/registration/embomeeting_reducedprice The conference will take place at the EMBL Advanced Training Centre, Heidelberg, Germany. ** ***** Reviewer's comments ***** Referee #1 (Comments on Novelty/Model System):

See remarks
Referee #1 (Other Remarks): I think that this is a very interesting study describing the cellular and molecular events featuring sterile inflammation in the eye. I particularly like the idea that the authors decided to work on an immunologically privileged organ like the eye since these tissues are profoundly immunosuppressive environments and hence ideal systems to study the effects of non-microbial damage associated molecular patterns.
The authors used both a chemical and mechanical injury as inflammatory stimuli and performed a detailed temporal scan of the signaling pathways molecules and cells involved in this process. The results showed that the inflammatory response could be divided in two phases with different inflammatory gene fingerprints and cellular players.
The experimental plan is very well thought and straightforward -the results clear and convincing. Nevertheless, I think that there are some points that deserve clarification or need to be addressed: 1. The authors described two waves of neutrophil infiltration in the cornea measuring the levels of MPO. Since this enzyme is also expressed by monocytes and macrophages, I wonder if these cells, which are commonly "late-comers" in inflammation, could also be present in the second phase of the response. Is it possible to perform some immunohistochemistry studies to show the phenotype of the cells infiltrated in the cornea? 2. Figure 1B. Is there any particular reason why the authors decided to focus their attention only on the genes that were upregulated? One might argue that the downregulation of certain homeostatic/protective genes might also play a role in the inflammatory response that has been described.
3. It is not clear to me why the authors did not measure the levels of secretoneurin (SN) in the cornea by ELISA (i.e. by a quantitative method). Also, I think that the authors should provide an explanation or discuss the possible cellular sources of the SN found in the serum of inflamed rats. Figure 3A are not very convincing. The difference in MPO levels following PBS or SN injections are very little (significant??) and comparable with those obtained following injection with HSPB4 -they both peak at 4hrs and reach the same level (8microgram/ml). It makes me wonder if this might be just an effect of injecting a recombinant protein. Along these lines, I also wonder how the authors decided to inject 0.2ng and if it would not be a better idea to test different doses. Finally, another protein of similar mw as SN should be used as negative control rather than PBS. Indeed, looking at the levels of MPO obtained 24 hr after injection of PBS in Figure 4A (about 30micrograms/ml and different from the ones showed in Fig. 3A and B) it does seems that the injection itself might cause some inflammation. 5. Figure 5B and C. Similar to the point above, a proper unrelated negative control should be used to show the specific anti-inflammatory effects of TSG-6 and an explanation for the dose that has been used provided.

The results in
6. Figure 5F. Although widely used in the past, the nuclear translocation of p65 is not a suitable assay to show NF-kB activation. In fact, recent studies have shown that NF-kB "oscillates" and shuttle back and forth between cytosol and nucleus during "inflammation" (i.e. following stimulation with inflammatory stimuli). 7. Page 9. Please change "dose-dependent manner" in "concentration-dependent manner". The results shown in Figure 5I and J do not support the hypothesis that "HSPB4 acted primarily through TLR2" -if anything, they suggest that HSPB4 activate both TLR2 and 4.

Referee #3 (Comments on Novelty/Model System):
Please see below.
Referee #3 (Other Remarks): The authors study the sterile inflammation of the cornea and demonstrate that it takes place in 2 phases associated with two waves of neutrophile recruitment. They then identify critical mediators of these 2 phases : the neuropeptide secretoneurin (SP) is a critical element of the early phase (acting in a macrophage-independent manner), while the small heat shock protein HSPB4 activates the TLR2/NF-kB axis in resident macrophages leading to the secretion of proinflammatory cytokines and recruitment of a second wave of neutrophils. TSG-6, a CD44-dependent inhibitor of the TLR2 axis, inhibits this second wave.
These results are important as they document the pathway involved in a relatively frequent pathology, and provide some potential ideas for therapy. Therefore they deserve publication, assuming the following points are adequately addressed : -The authors pick a single protein among 842 induced genes in phase 1. This choice is supported by 2 sets of data : 1) gain offunction exps involving injection of SP which stimulates the infiltration of neutrophils seen in phase I. 2) phase I is inhibited by application of a non-specific calcium channel blocker. Somme loss of function exps should be presented, such as usage of antibodies to SP, or other means, to substantiate the choice of this single potential mediator of phase I.
-To confirm the involvement of the TLR2 axis, it would be useful, again as a loss of function experiment, to study the situation in TLR2-/-mice.
-How does HSPB4 activate the TLR2 pathway? Does it directly bind TLR2 or does it act indirectly?
Minor point : is TSG-6 expression induced by HSPB4 stimulation, similar to a classical negative feedback loop? To Reviewer 1 I think that this is a very interesting study describing the cellular and molecular events featuring sterile inflammation in the eye. I particularly like the idea that the authors decided to work on an immunologically privileged organ like the eye since these tissues are profoundly immunosuppressive environments and hence ideal systems to study the effects of non-microbial damage associated molecular patterns.

The authors used both a chemical and mechanical injury as inflammatory stimuli and performed a detailed temporal scan of the signaling pathways molecules and cells involved in this process. The results showed that the inflammatory response could be divided in two phases with different inflammatory gene fingerprints and cellular players.
The experimental plan is very well thought and straightforward -the results clear and convincing. Nevertheless, I think that there are some points that deserve clarification or need to be addressed: 1. The authors described two waves of neutrophil infiltration in the cornea measuring the levels of MPO. Since this enzyme is also expressed by monocytes and macrophages, I wonder if these cells, which are commonly "late-comers" in inflammation, could also be present in the second phase of the response. Is it possible to perform some immunohistochemistry studies to show the phenotype of the cells infiltrated in the cornea? Response:  Having the same concern as the reviewer's, we had serially stained the corneal section at each time point after injury with neutrophil-specific marker (neutrophil elastase) as in our previous article (See Figure 6 of Proc Natl Acad Sci U S A. 2010 Sep 28;107(39):16875-80). We found that most of the cells infiltrated in the cornea were neutrophils until 48 hours of injury (most of the second phase of the inflammatory response). To address the reviewer's concern, we clarified this aspect in the Result section of the revised manuscript (Page 4, line 20-23).
2. Figure 1B. Is there any particular reason why the authors decided to focus their attention only on the genes that were upregulated? One might argue that the downregulation of certain homeostatic/protective genes might also play a role in the inflammatory response that has been described. Response:  In response to the reviewer's comment, we added to the revised Result and Discussion (Page 5, line 8-10 and Page 11, line 15-21) the explanation on why we evaluated the upregulated genes in the injured cornea as follows: "we screened for the molecules up-regulated early after an injury in order to search for the stimuli inducing Phase I and II responses under the assumption that the genes upregulated in the tissue in response to an injury might act as the major stimuli to initiate the sterile inflammation." 3. It is not clear to me why the authors did not measure the levels of secretoneurin (SN) in the cornea by ELISA (i.e. by a quantitative method). Also, I think that the authors should provide an explanation or discuss the possible cellular sources of the SN found in the serum of inflamed rats. Response:  We evaluated the levels of secretoneurin (SN) in the cornea by ELISA, which was the basis for determining the amount of SN to be injected into corneal stroma in Figure 3A. To address the reviewer's concern, we re-clarified this in the revised manuscript (Page 15, line 14-18).
Also, in response to the reviewer's comment, we added the rationale on why we measured the levels of SN in the serum (Page 6, line 13-15). Briefly, the neuropeptide secretoneurin is normally preformed and stored in sensory nerve endings of the cornea. Immediately after injury to the cornea, SN is released from the nerve endings. Thus, the measurement of SN in corneal extracts reflects the level of SN stored inside the nerves as well as the one secreted by injury. For that reason, we elected to measure the level of SN in the serum in order to evaluate the level of SN released outside of the corneal nerves by injury as a distinct form, not SN stored inside the nerve endings of the cornea ( Figure 2C). Also, we confirmed the increase in the SN levels in the cornea after injury using both western blotting and immunohistochemistry (Figure 2A and F).
As for the possible source of SN, we added the discussion to the Result section of the revised manuscript (Page 6, line 15-20). Figure 3A are not very convincing. Figure 4A (about 30micrograms/ml and different from the ones showed in Fig. 3A and B) it does seems that the injection itself might cause some inflammation.

Response:
 We agree with the reviewer that an intrastromal injection itself induced a certain level of inflammation as we could see in the PBS (phosphate-buffered saline) injection in Figures 3A and B. It is quite expected because an intracorneal injection is a type of mechanical damage to the cornea, although this inflammation would be very minor and clinically insignificant as the cataract surgeons do the intrastromal injection of fluid for wound closure at the end of every cataract surgery without any consecutive problems. To exclude the possibility that an increase in MPO might be a result of an intrastomal injection rather than the injected molecule, we used as a negative control an injection of the same volume of PBS (solvent of injected molecule) in every experiment ( Figures 3A and B). We used PBS because SN or HSPB4 was prepared in PBS to achieve the desired concentration. In fact, we had injected beta-crystallin (MW 23 kDa) into corneal stroma in one of our preliminary experiments and found no differences in MPO between beta-crystallin and PBS injected corneas. In contrast, as in Figure 3A, the differences in the MPO levels between SN-and PBS-injection groups were clearly and remarkably significant at all evaluated time-points. (To address the reviewer's concern, we clarified this in the revised manuscript by adding asterisks to denote the statistical significance to the graph of the revised Figure 3.) Hence, we do not think that the increase in MPO after SN injection was a nonspecific effect of recombinant protein injection. Moreover, when the release of SN was blocked from corneal nerves by instillation of the calcium-channel blocker, the Phase I inflammation was significantly suppressed in an injured cornea as in Figure 3D, which further supports that SN contributed to an early inflammation in the cornea.
We elected to inject 0.2 ng SN because it was the amount of SN detected in the cornea at 15 min after injury as we explained in the revised Method (Page 15, line 14-18). In preliminary experiment, we also injected 1 ng SN, as the reviewer suggested, which resulted in three times higher MPO compared to 0.2 ng SN injection. However, this dose (1 ng) was much higher than the actual level of endogenous SN (0.2 ng) observed in the injured cornea. For that reason, we did not include these data in the manuscript.
In regard to the differences in MPO between Figure 3B and Figure 4A, we attribute the differences to the fact that the injury method was different between the two experiments: For Figure 3A and B, we injected SN or HSPB4 into one site of the cornea, and thus the impact of the injury must have been localized. In contrast, in Figure 4A experiment, we injured the whole cornea by ethanol application/scraping of the whole corneal surface. These differences can be clearly seen in the histological pictures of the cornea in Figure 3C and Figure 4B: the inflammatory infiltration was localized around the injection site in Figure 3C, while there was a diffuse infiltration of inflammatory cells in Figure 4B. Figure 5B and C. Similar to the point above, a proper unrelated negative control should be used to show the specific anti-inflammatory effects of TSG-6 and an explanation for the dose that has been used provided.

5.
Response:  As negative control for the treatment group with an intracameral injection of TSG-6 (2 µg in 5 µL of PBS), we injected into the anterior chamber of the eye the same volume of PBS (5 µL) as indicated in Page 16, line 17-19. We used PBS because a lyophilized form of TSG-6 was prepared in PBS to achieve a desired concentration.
We used 2 µg of TSG-6 because this was the maximal and most effective dose as in the doseresponse curve from our previous experiment (See Figure  6. Figure 5F. Although widely used in the past, the nuclear translocation of p65 is not a suitable assay to show NF-kB activation. In fact, recent studies have shown that NF-kB "oscillates" and shuttle back and forth between cytosol and nucleus during "inflammation" (i.e. following stimulation with inflammatory stimuli). Response:  We agree with the reviewer that the staining of p65 at one time-point might not precisely reflect the NF-kB activation. Hence, in addition to the staining, we assayed for the NF-kB activation using the reporter cell line transfected to assay the TLR-mediated activation of the NF-ĸB signaling as in Figures 5H-J and 6C, D. This cell line is co-transfected with SEAP (secreted embryonic alkaline phosphatase) genes that is placed under the control of the IFN-β minimal promoter fused to five NF-κB and AP-1-binding sites. Thus, TLR stimulation and the activation of NF-κB and AP-1 can be determined in these cells by measuring the production of SEAP into the supernatant of the culture.
7. Page 9. Please change "dose-dependent manner" in "concentration-dependent manner". The results shown in Figure 5I and J do not support the hypothesis that "HSPB4 acted primarily through TLR2"-if anything, they suggest that HSPB4 activate both TLR2 and 4.

Response:
 Appropriate changes were made in the revised manuscript (Page 10, line 4 and 11; Page 34, line 8). In this study, we did most of the experiments to elucidate the effect of HSPB4 on TLR2/ NF-κB signaling in macrophages as a principal inducer of Phase II inflammation. We did this because we observed HSPB4 had a smaller effect in the reporter cell line that expressed TLR4, compared to TLR2-expressing cell line. Considering the reviewer's comment, we decided to delete the data on the effect of HSPB4 in TLR4-expressing cell line ( Figure 5I of the previous manuscript). Instead, per another reviewer's suggestion, we added the results of an additional experiment in which we confirmed that HSPB4 did not induce as much Phase II inflammation in TLR2 knockout mice as it did in wild-type mice ( Figure 5D of the revised manuscript). We appreciate both reviewers pointing out the important need to address this issue.
To Reviewer 3 The authors study the sterile inflammation of the cornea and demonstrate that it takes place in 2 phases associated with two waves of neutrophil recruitment. They then identify critical mediators of these 2 phases : the neuropeptide secretoneurin (SP) is a critical element of the early phase (acting in a macrophage-independent manner), while the small heat shock protein HSPB4 activates the TLR2/NF-kB axis in resident macrophages leading to the secretion of proinflammatory cytokines and recruitment of a second wave of neutrophils. TSG-6, a CD44-dependent inhibitor of the TLR2 axis, inhibits this second wave.
These results are important as they document the pathway involved in a relatively frequent pathology, and provide some potential ideas for therapy. Therefore they deserve publication, assuming the following points are adequately addressed: 1. The authors pick a single protein among 842 induced genes in phase 1. This choice is supported by 2 sets of data : 1) gain of function exps involving injection of SP which stimulates the infiltration of neutrophils seen in phase I. 2) phase I is inhibited by application of a non-specific calcium channel blocker. Some loss of function exps should be presented, such as usage of antibodies to SP, or other means, to substantiate the choice of this single potential mediator of phase I.

Response:
 As the loss of function experiment to clarify the role of the neuropeptide secretoneurin (SN), we blocked the release of SN from the sensory nerve endings of the cornea by instillation of the calcium-channel blocker (20 µl of 2 mM Diltiazem solution in isotonic saline) to the rat cornea 15 min before injury as in the previous article (Gonzalez et al., Invest Ophthalmol Vis Sci.1993;34:3329-333). We also confirmed that SN release was blocked by this method by observing that the level of SN was not detected in the serum after injury to the cornea that had Diltiazem instillation prior to injury. We clarified this in the revised manuscript (Page 8, line 5-7; Page 16, line 4-5).
2. To confirm the involvement of the TLR2 axis, it would be useful, again as a loss of function experiment, to study the situation in TLR2-/-mice.  TSG-6 as TLR2 antagonist Response:  First, we observed that TSG-6, that inhibits TLR2/NF-ĸB signaling (Choi et al., Blood 2010;118:330-8), suppressed Phase II but not Phase I in the cornea after sterile injury. Also, TSG-6 significantly suppressed the Phase II inflammation caused by an intrastromal injection of HSPB4. From these results, we noted that HSPB4 might act through TLR2/NF-ĸB signaling pathway, and we further confirmed these in a series of in vitro experiments using macrophages and TLR2expressing cell lines.
However, in agreement with the reviewer on that more direct animal experiment for a loss of function study is necessary, we additionally did the experiment using TLR2 knockout mice. We found that HSPB4 did not induce as much inflammation in Phase II in TLR2 knockout mice as it did in wild-type controls (B6). These results have been incorporated in the revised manuscript (Page 9, line 15-17; Figure 5D of the revised manuscript). We appreciate the reviewer pointing out the important need to address this issue.

How does HSPB4 activate the TLR2 pathway? Does it directly bind TLR2 or does it act indirectly?
Response:  Our experiments do not answer these questions. We note that a recent publication indicates that a functional interaction between heat shock protein 90 and Beclin 1 is required for TLR signaling under some conditions (Xu et al., 2011. FASEB J 25:2700. Therefore, as the reviewer suggests, the interaction between HSPB4 and TLR2 may be complex. We respectfully submit investigating this interaction may be beyond the scope of the present manuscript.

Minor point: is TSG-6 expression induced by HSPB4 stimulation, similar to a classical negative feedback loop?
Response:  From our experiment, the expression of TSG-6 gene (Tnfip6) in the cornea was increased up to 4.2 and 11.2 times of the baseline levels at 4 and 24 hours after HSPB4 100 ng injection, respectively, as assayed by real-time RT-PCR. Hence, we believe that the cornea or other tissues increase the expression of TSG-6 in response to an injury by HSPB4 or other stimuli in a classical negative feedback loop as the reviewer indicated. Thank you for the submission of your revised manuscript "Identification of the HSPB4/TLR2/NF κB axis in macrophage as a therapeutic target for sterile inflammation of the cornea" to EMBO Molecular Medicine. We have now received the report from the reviewer who was asked to rereview your manuscript. You will be glad to see that the reviewer is now globally supportive and we can proceed with official acceptance of your manuscript pending the changes detailed below.
As highlighted by the reviewer, please include the data you mention in the point-by-point response regarding the recombinant protein control and the use of higher amounts of SN in the