Interferon‐beta signaling in retinal mononuclear phagocytes attenuates pathological neovascularization

Abstract Age‐related macular degeneration (AMD) is a leading cause of vision loss among the elderly. AMD pathogenesis involves chronic activation of the innate immune system including complement factors and microglia/macrophage reactivity in the retina. Here, we show that lack of interferon‐β signaling in the retina accelerates mononuclear phagocyte reactivity and promotes choroidal neovascularization (CNV) in the laser model of neovascular AMD. Complete deletion of interferon‐α/β receptor (Ifnar) using Ifnar1−/− mice significantly enhanced early microglia and macrophage activation in lesion areas. This triggered subsequent vascular leakage and CNV at later stages. Similar findings were obtained in laser‐treated Cx3cr1 Cre ER:Ifnar1 fl/fl animals that allowed the tamoxifen‐induced conditional depletion of Ifnar in resident mononuclear phagocytes only. Conversely, systemic IFN‐β therapy of laser‐treated wild‐type animals effectively attenuated microgliosis and macrophage responses in the early stage of disease and significantly reduced CNV size in the late phase. Our results reveal a protective role of Ifnar signaling in retinal immune homeostasis and highlight a potential use for IFN‐β therapy in the eye to limit chronic inflammation and pathological angiogenesis in AMD.

Thank you for the submission of your manuscript to EMBO Molecular Medicine. We have now heard back from the three referees whom we asked to evaluate your manuscript. Although the referees find the study to be of potential interest, they also raise a number of concerns that must be addressed in the next final version of your article.
You will see from the comments below that all referees find the study interesting and referee 3 particularly highlights its translational relevance. While referees 2 and 3 are generally more supportive, referee 1 is rather critical of the experimental design and is concerned about the CNV model and the mouse model used as well as existing literature previously establishing that IFNbeta inhibits laser-induced CNV in rabbits and monkeys. As we feel that requested experiments would certainly improve the robustness of the data, we would like to invite you to address all issues as suggested, and experimentally when needed.
Given these evaluations, I would like to give you the opportunity to revise your manuscript, with the understanding that the referee concerns must be fully addressed and that acceptance of the manuscript would entail a second round of review. Please note that it is EMBO Molecular Medicine policy to allow only a single round of revision and that, as acceptance or rejection of the manuscript will depend on another round of review, your responses should be as complete as possible.
I look forward to seeing a revised form of your manuscript as soon as possible.

Introduction :
The privious study by Yasukawa et al on interferon and CNV in rabbits is not mentioned in the instroduction. A previous study on laser-induced CNV in monkeys is not cited anywhere in the manuscript (Nippon Ganka Gakkai Zasshi. 1995 May;99(5):571-81.The effect of interferon-beta on experimental choroidal neovascularization. Tobe T1, Takahashi K, Ohkuma H, Uyama M.) There also is an interesting case report about an interferon beta and CNV that is not cited at all (Retina.  Figure 1 the authors analyzed IBA1+cell infiltration (d3) and CNV (d3, d7, d14) in wildtype and Ifnar-/-mice. The results presented in figure 1 do not allow a diferentiation between microglia, infiltrating inflammatory macrophages or choroidal resident macrophages. The authors need to be more precise in their wording as the cells they count in the lesions likely are a mixture of all three cell types. The authors counted amoboid IBA1+cells and crossing points. The total number of IBA1+cells needs also to be shown. Why were the IBA1+ cells only quantified at d3, when the vascular changes are only apparant at d14 ? A later analysis of IBA1 cell numbers and phenotype is necessary if the authors think that their the IBA1 cell phenotype is the reason for the vascular differences. The CNV lesions are very close to the optic nerve. In panel N and P, CNV bridging two to three laser lesions are visible. When laser lesions are two clos to each other a CNV lesion often forms between two laser lesions, which is bigger than the addition of two seperate lesions. I would therefore suggest not to include bridging CNV lesions, as their size is also influenced by the aleatory distance to the neighbouring CNV. Also I am not sure I understand the quantification method of angiography fluorescence by choosing two regions of interest (ROI) within and one ROI outside the laser spots. Are the CNV size quantifications per/Impact ? if so, how were confluent CNVs quantified ? Why did the authors use lectin as a vascular stain, as it also marks activated IBA1 cells? Figure 2. : In the experiments represented in Figure 2 the authors analyzed IBA1+cell infiltration (d3) and CNV (d3, d7, d14) in mice treated with IFN beta. Again the IBA1 cell number and phenotypes were only analyzed at d3, when the vascular differences only appeared 11days later. Also the total number of IBA1+cells is again missing. Panel N reveals confluent CNVs. Also the laser spots are sometimes distributed equally around the optic nerve (M) sometimes only to one side of the ON(N). The authors also measured ´ edema ª formation in this set of experiments. Edema is a fluid accumulation within the tissue (either intracellularly or inter cellularly). How do the authors distinguish between edema and infiltrating cells (IBA1+cells and proliferating endothelial cells) in the OCT images ?  Figure 3 the authors analyzed IBA1+cell infiltration (d3) and CNV (d3, d7, d14) in Ifnar1fl/fl mice and tamoxifen-treated Cx3cr1CreER:Ifnar1fl/fl mice. The authors induced Cre expression 28 and 26 days prior to the laser lesion which should permanently delete the Ifnar1 gene in cells that express high levels of Cx3cr1. 26 days after the TAM treatment cells with a high turnover, such as monocytes will again express Ifnar. The authors therefore state that the experimental mice only have a lack of Ifnar in microglial cells. This is however likely incorrect, as resident macrophages in choroid and ciliary body also have a very slow turnover. They likely participate in the laser-induced inflammation (they are actually closer to the burn than microglial cells) and will likely still lack (at least in part) Ifnar expression. It is therefore nor possibly to decipher the role of ifnar in only microglial cells using these mice. The efficiency of Ifnar deletion in Cx3cr1 expressing cells with high and low turnover were not analyzed in the hands of the authors and no data from Cx3cr1 cre expressing mice is presented. The Iba1 cells in panel B (Ifnar1 fl/fl) look very differently to the ones shown in Fig. 1B (wildtype) why is that ? The laser spots are again very close to the optic nerve (one CNV in panel N seems to actually merge with the optic nerve). They are again not distributed equally, sometimes being only on one side of the optic nerve, sometimes all around.
Referee #2 (Comments on Novelty/Model System): Due to lack of animal models that approximate macular degeneration, laser injury may be acceptable.
Referee #2 (Remarks): mboThe goal of the study was to analyze the role of interferon (IFN)-fl and its receptor (IFNAR) in the laser model of retinal injury The key findings of this study were: (1) Laser-treated Ifnar1-/-mice showed enhanced microglia activity resulting in vascular leakage and choroidal neovascularization (CNV). (2) Laser-treated Cx3cr1CreER:Ifnar1fl/fl mice (ie depletion of IFNAR in microglia only) showed similar effects. (3) IFN-fl treated wild-type mice displayed decreased microgliosis, vascular leakage and CNV lesion size. This results demonstrate a strong influence of IFN-fl signaling on retinal microglia activity and may offer new therapeutic strategies for acute retinal injuries.

Major comments
1. The authors should show cross sections of the CNV lesions induced laser coagulation as well as cross sections stained with Iba1. 2. It would be informative to include the expression level of Ifnar in wt, Ifnar1-/-and Cx3cr1CreER:Ifnar1fl/fl mice in the retina using immunoblot and immunohistochemistry 3. Acute laser injury may not have relevance to a chronic degeneration disorder, including age related macular degeneration.
1. Laser-treated wild-type mice are labeled as C57BL6/J in Figure 1 and as control in Figure 2 please make it consistent. 2. Figure 2C: the label "+IFN-fl" is not necessary as it Referee #3 (Comments on Novelty/Model System): This is an exciting study with significant translational potential. The model systems are adequate but some additonal controls would be required especially in the studies using tamoxifen induced Cre recombinase expression in microglial cells. Further details are outline din my comments below.

Referee #3 (Remarks):
This manuscript by L¸ckoff and colleagues focuses on the role of interferon-beta in the regulation of choroidal neovascularisation, the end stage of the relatively common form of blindness wet agerelated macular degeneration (AMD). The authors also show that interferon-beta is intimately involved in regulating microglial homeostasis and a lack of IFN-beta can induce microgliosis.
Using mice lacking the interferon receptor, the authors show that microglial reactivity was increased in regions of laser induced CNV (a model of neovascular AMD). The authors showed similar results in floxed mice that had IFNAR deleted specifically in microglial cells.
Strikingly, and what represents an exciting translational finding, the authors showed that systemic IFN-beta therapy in mice post induction of laser CNV showed significantly decreased CNV lesions and that IFN-beta therapy could have utility in the treatment of neovascular AMD patients.
Below are some specific comments for the authors to address 1. The quantification of amoeboid shaped cells in Fig 1B/D will need to be elaborated upon in the methods section as it strikes me as a very subjective way of data analysis.
2. What is the homology of human IFN-beta to mouse IFN-beta and are there differences in bioactivity between the two? Human IFN-beta only has 60% homology to mouse IFN-beta...would there be differences in therapeutic readout if the authors used mouse IFN-beta? 3. Figure 3 should have included a control group of the Cx3cr1CreER mice on their own. It is widely accepted that Cre recombinase can have biological effects and toxicity when expressed in cells and this control would be important to include in the figure. This control would need to be used in all sub-sections of Figure 3. It will markedly strengthen the paper. There are however several major problems that should be addressed.
1. Generally the laser lesions are (i) too close to the optic nerve (they are classically applied 4-5 optic nerve diameters away from the optic nerve) and (ii) not evenly distributed around the optic nerve. As a result some of the CNV formations fuse between two laser lesions which greatly influences their size as the bridging CNV lesions are bigger than two single CNV lesions.

Response:
When studying laser-coagulation in mice there are several limitations mainly due to the small eye, large lens and the differences in optics of small rodents. We have done our best to ensure that the laser lesions were equally distributed at 3, 9 and

Response to reviewers Figure: Representative infrared fundus images of laser coagulation experiments for all different mouse strains and the treatment study presented in our manuscript.
Relatively equal distributions without fusions can be noticed.
2. The inflammation was only analyzed at d3, even though the vascular changes are invariably only observed at d14. If the differences in IBA1+cells at d3 were responsible for the vascular phenotype, you would expect the vascular phenotype to be different at d7. It seems therefore likely that the IFNAR1-dependent differences in inflammatory phenotype that are responsible for the vascular changes occur at a later stage, or they should at least be analyzed additionally at a later time point.

Response:
We fully agree with the reviewer. 3. The analysis of the Cx3cr1CreER:Ifnar1fl/fl mice does not allow to specifically analyze microglial IFNAR1 expression per se in this model, as the lesion was made in the choroid that contains numerous resident macrophages that likely express cx3cr1 and also have a slow turnover.

Response:
We agree with the reviewer that potentially long-lived resident Iba1+ cells e.g. choroidal macrophages could also contribute to the effect seen when using the Cx3cr1 CreER :Ifnar1 flox/flox model. To overcome the potential confusion when using only the term "microglia", we have now throughout the manuscript used the terms "Iba1+ cells", "microgia/macrophages" or "mononuclear phagocytes". We think that all three terms are scientifically correct and we have also included a reference that gives some information on non-microglial mononuclear cells in the eye: McMenamin PG (1999) Dendritic cells and macrophages in the uveal tract of the normal mouse eye. Br J Ophthalmol 83:598-604. However, this reference and also other papers we found did not really define how long these cells could potentially live or when they are repopulated.

Also the deletion of Ifnar1 was not controlled for in the tissue and the different macrophage populations in the hands of the authors and a Cre-expressing control is missing.
Response: We have now performed additional experiments to demonstrate Ifnar deletion in Ifnar1 -/and Cx3cr1 CreER :Ifnar1 flox/flox animals (Appendix Supplementary Figure S3). Therefore, we have used genomic PCR to demonstrate the genomic deletion of Ifnar1 exon 10 (Appendix Fig.S3A To overcome the potential confusion when using only the term "microglia", we have now throughout the manuscript used the terms "Iba1+ cells", "microglia/macrophages" or "mononuclear phagocytes".

Response:
The publication by Yasukawa et al. was already cited in the first submitted manuscript but was included in the discussion and not introduction. Since the Yasukawa article is well suited to discuss common and divergent findings related to our study we still think that discussion is the better place than introduction. We

Response: Please see response to question 3
To overcome the potential confusion when using only the term "microglia", we have now throughout the manuscript used the terms "Iba1+ cells", "microglia/macrophages" or "mononuclear phagocytes".
9. The authors counted amoeboid IBA1+cells and crossing points. The total number of IBA1+cells needs also to be shown.

Response:
The total number of Iba1+ cells was counted and is now included in Appendix Figure S1. It was not statistically different between the mouse groups analyzed.
10. Why were the IBA1+ cells only quantified at d3, when the vascular changes are only apparent at d14? A later analysis of IBA1 cell numbers and phenotype is necessary if the authors think that the Iba1 cell phenotype is the reason for the vascular differences.
Response: Please see response to question 2

The CNV lesions are very close to the optic nerve. In panel N and P, CNV bridging two to three laser lesions are visible. When laser lesions are two close to each other a CNV lesion often forms between two laser lesions, which is bigger than the addition of two separate lesions. I would therefore suggest not to include bridging CNV lesions, as their size is also influenced by the aleatory distance to the neighboring CNV.
Response: Please see response to question 1

Response The CNV quantification was done as mean values per eye. Two ROIs inside the leakage and one lesion in the background are illustrated in the next figure (for review purpose only) (A). To localize the laser spots, infrared images taken in the same positions were used as reference (B).
Response to reviewers Figure: Representative fundus fluorescein angiography and infrared fundus images of laser coagulation experiments delineating the ROI and quantification methods.

The text in the Materials and Methods section reads as follows: "To quantify the laser-induced vascular leakage the pixel intensity was measured in two regions of interests (ROI) within and one ROI outside each laser spot using the image processing program ImageJ (NIH). Because three laser spots were induced per eye, we quantified the pixel intensity of six ROI within and three ROI outside the fluorescein leakages. After averaging the values and subtracting the background, one data point represented the mean laser-induced leakage per eye. Eyes were excluded from the analysis in case of choroidal hemorrhages and when laser lesions
completely fused with each other or the optic nerve head."

Why did the authors use lectin as a vascular stain, as it also marks activated IBA1 cells?
Response Iba1-lectin co-staining of retinal cryosections and RPE/choroidal flat mounts revealed that lectin does not co-stain mononuclear phagocytes similarly as with Iba1 (see Expanded View Figures EV1, EV2, EV3). Staining with lectin is a commonly used and widely cited technique in the field to detect neovessels and there is no superiority of FITC-dextran staining in our opinion. Moreover, dextranstaining requires an elaborate animal perfusion to reach all blood vessels. This procedure is quite error-prone and not easy to standardize. For the reviewer, we have performed dextran/lectin double stainings of mouse RPE/choroidal flat mounts 7 days after laser coagulation. This image shows that lectin is well suited and that it seems superior to dextran as it stains more vessels and can be better quantified.

Response to reviewers Figure: Representative co-labeling of lectin and dextran in mouse RPE/choroidal flat mounts 7 days after laser coagulation.
14. Figure 2: In the experiments represented in Figure 2 the authors analyzed IBA1+cell infiltration (d3) and CNV (d3, d7, d14) in mice treated with IFN beta. Again the IBA1 cell number and phenotypes were only analyzed at d3, when the vascular differences only appeared 11days later.

Also the total number of IBA1+cells is again missing.
Response: Please see response to question 9

Panel N reveals confluent CNVs. Also the laser spots are sometimes distributed equally around the optic nerve (M) sometimes only to one side of the ON (N).
Response: Please see response to question 1

The authors also measured &#x00AB; edema &#x00BB; formation in this set of experiments. Edema is a fluid accumulation within the tissue (either intracellular or inter cellular). How do the authors distinguish between edema and infiltrating cells (IBA1+cells and proliferating endothelial cells) in the OCT images?
Response: Since we do not have ultimate proof that the structure seen in OCT is edema/fluid and since these data are not fundamental for the story, we have excluded this figure subpanel from the revised paper.

days after the TAM treatment cells with a high turnover, such as monocytes will again express Ifnar. The authors therefore state that the experimental mice only have a lack of Ifnar in microglial cells. This is however likely incorrect, as resident macrophages in choroid and ciliary body also have a very slow turnover. They likely participate in the laser-induced inflammation (they are actually closer to the burn than microglial cells) and will likely still lack (at least in part) Ifnar expression. It is therefore nor possibly to decipher the role of ifnar in only microglial cells using these mice.
Response: Please see response to question 3

19.The efficiency of Ifnar deletion in Cx3cr1 expressing cells with high and low turnover were not analyzed in the hands of the authors and no data from Cx3cr1 cre expressing mice is presented.
Response: Please see response to question 4 21. The laser spots are again very close to the optic nerve (one CNV in panel N seems to actually merge with the optic nerve). They are again not distributed equally, sometimes being only on one side of the optic nerve, sometimes all around.

The authors should show cross sections of the CNV lesions induced laser coagulation as well as cross sections stained with Iba1.
Response: We fully agree with this comment that partially relates to question 4 of reviewer 1. We have now performed additional experiments to demonstrate Ifnar deletion in Ifnar1 -/and Cx3cr1 CreER :Ifnar1 flox/flox animals (Appendix Supplementary Figure S3). Therefore, we have used genomic PCR to demonstrate the genomic deletion of Ifnar1 exon 10 (Appendix Fig.S3A), Western blot analysis of total retinal extracts of Ifnar1 -/and Cx3cr1 CreER :Ifnar1 flox/flox animals using a specific anti-Ifnar antibody (Appendix Fig.S3B), and immunohistochemical staining of sections from Ifnar1 -/and Cx3cr1 CreER :Ifnar1 flox/flox mice using the same specific anti-Ifnar antibody together with Iba1 (Appendix Fig.S3A). We also tried to perform ex vivo isolation of Iba1+ cells with MACS and thereafter perform FACS analysis but this experimental set up repeatedly failed because of limitations in total Iba1+ cell numbers and obviously incompatibility of the antibody for FACS.

Response:
We are aware of the fact that the murine laser-coagulation model has some limitations especially related to the aspect of aging. However, our hypothesis for this work was that Ifnar signaling has a potential therapeutic effect by limiting retinal inflammation and thereby indirectly also choroidal neovascularization, both typical hallmarks of AMD. These aspects were well covered by the model and in addition we had the option to use several different genetically modified animals. Throughout the manuscript we were very careful not to over-interpret our findings.
Minor comments: 1. Laser-treated wild-type mice are labeled as C57BL6/J in Figure 1 and as control in Figure 2 -please make it consistent.

Response:
The labels in Figure 2 were changed to 'C57BL6/J'.

Figure 2C: the label "+IFN-ß is not necessary as it
Response: The label "+IFN-ß" in Figure 2C was removed.

The quantification of amoeboid shaped cells in Fig 1B/D will need to be elaborated upon in the methods section as it strikes me as a very subjective way of data analysis.
Response: For quantification of mononuclear phagocyte morphology, we followed a method using   Figure 3. It will markedly strengthen the paper.

Response:
This comment was very valid and we have now included the analysis of Cx3cr1 CreER mice in all subpanels of Figure 3 as well as in Figure EV3. Moreover, the text was changed to include this important control at all relevant passages of the manuscript. Thank you for the submission of your revised manuscript to EMBO Molecular Medicine. We have now received the enclosed reports from the referees that were asked to re-assess it. As you will see the reviewers are now supportive and I am pleased to inform you that we will be able to accept your manuscript pending final editorial amendments.
I look forward to receiving a new revised version of your manuscript as soon as possible, and within 2 weeks.
***** Reviewer's comments ***** Referee #1 (Comments on Novelty/Model System): while the effect of IFN beta on CNV was previously known this paper shows its important role on mononuclear phagocytes using adequate genetic mouse models. I am not sure this results will directly translate into a new medical approach as the major pharmacological culprits of IFN stability and necessary slow release formulations seem to me to be a remaining major challenge.
Referee #1 (Remarks): The additional experiments (analysis at intermediate time points, the verification of gene deletion and the analysis of Cre expressing controls) and the correct wording ("Iba1+ cells", "microgia/macrophages" or "mononuclear phagocytes") in the revised manuscript respond to all my concerns . I think this is a very nice study that highlights the importance of IFN signaling in resident mononuclear phagocytes. I would still advise the authors to place their laser impacts at greater distance from each other and the optic nerve in their futur studies (it is feasible in mice, we do it all the time).
Referee #2 (Comments on Novelty/Model System): Due to lack of animal models that approximate macular degeneration, laser injury may be acceptable.

Referee #3 (Comments on Novelty/Model System):
This manuscript has been substantially improved and authors have addressed comments in detail. 3. Were any steps taken to minimize the effects of subjective bias when allocating animals/samples to treatment (e.g. randomization procedure)? If yes, please describe.
For animal studies, include a statement about randomization even if no randomization was used.
4.a. Were any steps taken to minimize the effects of subjective bias during group allocation or/and when assessing results (e.g. blinding of the investigator)? If yes please describe. Do the data meet the assumptions of the tests (e.g., normal distribution)? Describe any methods used to assess it.
Is there an estimate of variation within each group of data?
To verify the statistical tests the Institute of Medical Statistics, Informatics and Epidemiology (IMSIE), University of Cologne, Cologne, Germany was consulted before applying the analysis.
The significance of the data was tested with one--way ANOVA. A Tukey's multiple comparison test was used and p--values greater that 0.05 were defined significant. For further details see method section of the manuscript.
There was normal variation in the data as every single animal reacts individually to the laser damage or the respective treatment. However, the variation within the experiments was reduced by using disease--free animals of similar weight and age.

YOU MUST COMPLETE ALL CELLS WITH A PINK BACKGROUND 
To determine the sample size in all animal experiments we performed a power analysis using G--POWER software. The effect sizes of the respective methods we experienced from pilot experiments or earlier publications. The animals were treated with IFN--ß or tamoxifen cage--wise. The cages were randomly allocated to the experimental groups.
A randomization was used when treating the animals with IFN--ß or tamoxifen, respectively.
The animals had consecutive numbers which were allocated to the genotype/treatment status only after cpmplete experimental evaluation. Also the staining procedure was performed using the consecutive numbering which was only unrevealed after analysis.

Data
the data were obtained and processed according to the field's best practice and are presented to reflect the results of the experiments in an accurate and unbiased manner. figure panels include only data points, measurements or observations that can be compared to each other in a scientifically meaningful way. graphs include clearly labeled error bars for independent experiments and sample sizes. Unless justified, error bars should not be shown for technical replicates. if n< 5, the individual data points from each experiment should be plotted and any statistical test employed should be justified Please fill out these boxes  (Do not worry if you cannot see all your text once you press return) a specification of the experimental system investigated (eg cell line, species name).
Each figure caption should contain the following information, for each panel where they are relevant:

Captions
The data shown in figures should satisfy the following conditions: Source Data should be included to report the data underlying graphs. Please follow the guidelines set out in the author ship guidelines on Data Presentation. a statement of how many times the experiment shown was independently replicated in the laboratory.
Any descriptions too long for the figure legend should be included in the methods section and/or with the source data.
Please ensure that the answers to the following questions are reported in the manuscript itself. We encourage you to include a specific subsection in the methods section for statistics, reagents, animal models and human subjects.
In the pink boxes below, provide the page number(s) of the manuscript draft or figure legend(s) where the information can be located. Every question should be answered. If the question is not relevant to your research, please write NA (non applicable).

B--Statistics and general methods
the assay(s) and method(s) used to carry out the reported observations and measurements an explicit mention of the biological and chemical entity(ies) that are being measured. an explicit mention of the biological and chemical entity(ies) that are altered/varied/perturbed in a controlled manner. the exact sample size (n) for each experimental group/condition, given as a number, not a range; a description of the sample collection allowing the reader to understand whether the samples represent technical or biological replicates (including how many animals, litters, cultures, etc.).

Reporting Checklist For Life Sciences Articles (Rev. July 2015)
This checklist is used to ensure good reporting standards and to improve the reproducibility of published results. These guidelines are consistent with the Principles and Guidelines for Reporting Preclinical Research issued by the NIH in 2014. Please follow the journal's authorship guidelines in preparing your manuscript.